back to indexElon Musk: SpaceX, Mars, Tesla Autopilot, Self-Driving, Robotics, and AI | Lex Fridman Podcast #252
link |
The following is a conversation with Elon Musk.
link |
His third time on this, the Lex Friedman podcast.
link |
Yeah, make yourself comfortable.
link |
Do you don't do the headphones thing?
link |
I mean, how close do I get?
link |
I need to get to this thing.
link |
The closer you are, the sexier you sound.
link |
Can't get enough of the all that, baby.
link |
I'm gonna clip that out anytime somebody messaged me about it.
link |
Yeah, you want my body.
link |
And you think I'm sexy.
link |
Come right out and tell me so.
link |
Do, do, do, do, do.
link |
Serious mode activate.
link |
Come on, you're Russian, you can be serious.
link |
Everyone's serious all the time in Russia.
link |
Allow me to say that the SpaceX launch
link |
of human beings to orbit on May 30th, 2020
link |
was seen by many as the first step
link |
in a new era of human space exploration.
link |
These human spaceflight missions were a beacon of hope
link |
to me and to millions over the past two years
link |
as our world has been going through
link |
one of the most difficult periods in recent human history.
link |
We saw, we see the rise of division, fear, cynicism,
link |
and the loss of common humanity,
link |
right when it is needed most.
link |
So first, Elon, let me say thank you
link |
for giving the world hope and reason
link |
to be excited about the future.
link |
Oh, it's kind of you to say.
link |
I do want to do that.
link |
Humanity has obviously a lot of issues
link |
and people at times do bad things,
link |
but despite all that, I love humanity
link |
and I think we should make sure we do everything
link |
we can to have a good future and an exciting future
link |
and one where that maximizes the happiness of the people.
link |
Let me ask about Crew Dragon demo two.
link |
So that first flight with humans on board,
link |
how did you feel leading up to that launch?
link |
Was it going through your mind?
link |
So much was at stake.
link |
Yeah, no, that was extremely stressful, no question.
link |
We obviously could not let them down in any way.
link |
So extremely stressful, I'd say, to say the least.
link |
I was confident that at the time that we launched,
link |
that no one could think of anything at all to do
link |
that would improve the probability of success
link |
and we racked our brains to think of any possible way
link |
to improve the probability of success.
link |
We could not think of anything more and nor could NASA
link |
and so that's just the best that we could do.
link |
So then we went ahead and launched.
link |
Now, I'm not a religious person, but I nonetheless
link |
got on my knees and prayed for that mission.
link |
Were you able to sleep?
link |
How did it feel when it was a success?
link |
First, when the launch was a success
link |
and when they returned back home or back to Earth?
link |
It was a great relief.
link |
Yeah, for high stress situations,
link |
I find it's not so much elation as relief.
link |
And I think once, as we got more comfortable and proved out
link |
the systems, because we really got to make sure everything
link |
works, it was definitely a lot more enjoyable
link |
with the subsequent astronaut missions.
link |
And I thought the inspiration mission was actually
link |
very inspiring, the inspiration for mission.
link |
I'd encourage people to watch the inspiration
link |
documentary on Netflix.
link |
It's actually really good.
link |
And it really isn't.
link |
I was actually inspired by that.
link |
And so that one, I felt I was able to enjoy the actual mission
link |
and not just be super stressed all the time.
link |
So for people that somehow don't know,
link |
it's the all civilian first time, all civilian out
link |
to space, out to orbit.
link |
Yeah, and it was, I think, the highest orbit
link |
that in like 30 or 40 years or something.
link |
The only one that was higher was the one shuttle, sorry,
link |
Hubble servicing mission.
link |
And then before that, it would have been Apollo in 72.
link |
And I think as a species, we want
link |
to be continuing to do better and reach higher ground.
link |
And I think it would be extremely tragic
link |
if Apollo was the high watermark for humanity.
link |
And that's as far as we ever got.
link |
And it's concerning that here we are, 49 years
link |
after the last mission to the moon.
link |
And so almost half a century, and we've not been back.
link |
And that's worrying.
link |
It's like, does that mean we've peaked as a civilization
link |
So I think we've got to get back to the moon
link |
and build a base there, a science base.
link |
I think we could learn a lot about the nature of the universe
link |
if we have a proper science base on the moon.
link |
Like we have a science base in Antarctica
link |
and many other parts of the world.
link |
And so that's the next big thing.
link |
We've got to have a serious moon base
link |
and then get people to Mars and get out there
link |
and be a space bearing civilization.
link |
I'll ask you about some of those details.
link |
But since you're so busy with the hard engineering
link |
challenges of everything that's involved,
link |
are you still able to marvel at the magic of it all,
link |
of space travel, of every time the rocket goes up,
link |
especially when it's a crewed mission?
link |
Or are you just so overwhelmed with all the challenges
link |
that you have to solve?
link |
And actually, to add to that, the reason
link |
that I wanted to ask this question of May 30th,
link |
it's been some time.
link |
So you can look back and think about the impact already.
link |
It's already, at the time, it was an engineering problem,
link |
Now it's becoming a historic moment.
link |
Like it's a moment that, how many moments
link |
would be remembered about the 21st century?
link |
To me, that or something like that,
link |
maybe inspiration for one of those
link |
would be remembered as the early steps of a new age
link |
of space exploration.
link |
Yeah, I mean, during the launches itself,
link |
so I think maybe some people will know,
link |
but a lot of people don't know.
link |
It's like, I'm actually the chief engineer of SpaceX.
link |
So I've signed off on pretty much all the design decisions.
link |
And so if there's something that goes wrong with that,
link |
a vehicle, it's fundamentally my fault.
link |
So I'm really just thinking about all the things that.
link |
So when I see the rocket, I see all the things
link |
that could go wrong and the things that could be better.
link |
And the same with the Dragon spacecraft,
link |
it's like, a lot of people will say, oh, this
link |
is a spacecraft or a rocket.
link |
And this looks really cool.
link |
I'm like, I've like a readout of like, these are the risks.
link |
These are the problems.
link |
That's what I see.
link |
Like, tch, tch, tch, tch, tch, tch.
link |
So it's not what other people see when they see the product.
link |
So let me ask you then to analyze Starship in that same way.
link |
I know you'll talk about it in more detail about Starship
link |
in the near future, perhaps.
link |
Yeah, we'll talk about it now if you want.
link |
But just in that same way, like you said,
link |
you see when you see a rocket, you see a list of risks.
link |
In that same way, you said that Starship
link |
was a really hard problem.
link |
So there's many ways I can ask this.
link |
But if you magically could solve one problem perfectly,
link |
one engineering problem perfectly, which one would it be?
link |
Sorry, on Starship.
link |
So is it maybe related to the efficiency, the engine,
link |
the weight of the different components, the complexity
link |
of various things, maybe the controls of the crazy thing
link |
as to do the land?
link |
Oh, actually, by far, the biggest thing absorbing my time
link |
is engine production, not the design of the engine.
link |
I've often said prototypes are easy, production is hard.
link |
So we have the most advanced rocket engine
link |
that's ever been designed.
link |
Because I say currently the best rocket engine ever
link |
is probably the RD180 or RD170, the Dora Russian engine,
link |
And still, I think an engine should only count
link |
if it's gotten something to orbit.
link |
So our engine has not gotten anything to orbit yet.
link |
It's the first engine that's actually
link |
better than the Russian RD engines,
link |
which were an amazing design.
link |
So you're talking about Raptor engine.
link |
What makes it amazing?
link |
What are the different aspects of it that make it?
link |
What do you get the most excited about
link |
if the whole thing works in terms of efficiency,
link |
all those kinds of things?
link |
Well, it's, but Raptor is a full flow staged combustion
link |
engine, and it's operating at a very high chamber pressure.
link |
So one of the key figures perhaps the key figure of merit
link |
is what is the chamber pressure at which the rocket engine
link |
That's the combustion chamber pressure.
link |
So Raptor is designed to operate at 300 bar, possibly
link |
maybe higher, 300 atmospheres.
link |
So the record right now for operational engine
link |
is the RD engine that I mentioned, the Russian RD, which
link |
is, I believe, around 267 bar.
link |
And the difficulty of the chamber pressure
link |
is increases on a nonlinear basis.
link |
So 10% more chamber pressure is more like 50%, more difficult.
link |
But that chamber pressure, that is
link |
what allows you to get a very high power
link |
density for the engine.
link |
So enabling a very high thrust to weight ratio
link |
and a very high specific impulse.
link |
So specific impulse is like a measure
link |
of the efficiency of a rocket engine.
link |
It's really the effect of exhaust
link |
velocity of the gas coming out of the engine.
link |
So with a very high chamber pressure,
link |
you can have a compact engine that nonetheless
link |
has a high expansion ratio, which
link |
is the ratio between the exit nozzle and the throat.
link |
So you see a rocket engine's got sort of like a hourglass shape.
link |
It's like a chamber and then it necks down and there's a nozzle.
link |
And the ratio of the exit diameter
link |
to the throat is the expansion ratio.
link |
So why is it such a hard engine to manufacture at scale?
link |
It's very complex.
link |
So what is complexity mean here?
link |
There's a lot of components involved.
link |
There's a lot of components and a lot of unique materials.
link |
So we had to invent several alloys
link |
that don't exist in order to make this engine work.
link |
So it's a materials problem, too.
link |
It's a materials problem.
link |
And in a full flow stage combustion,
link |
there are many feedback loops in the system.
link |
So basically, you've got propellants and hot gas
link |
flowing simultaneously to so many different places on the engine.
link |
And they all have a recursive effect on each other.
link |
So you change one thing here.
link |
It has a recursive effect here.
link |
It changes something over there.
link |
And it's quite hard to control.
link |
Like there's a reason no one's made this before.
link |
But the reason we're doing a stage combustion full flow
link |
is because it has the highest theoretical possible efficiency.
link |
So in order to make a fully reusable rocket,
link |
that's really the holy grail of orbital rocketry.
link |
You have to have everything's got to be the best.
link |
It's got to be the best engine, the best airframe,
link |
the best heat shield, extremely light avionics,
link |
very clever control mechanisms.
link |
You've got to shed mass in any possible way that you can.
link |
For example, instead of putting landing legs on the booster
link |
and ship, we are going to catch them with a tower
link |
to save the weight of the landing legs.
link |
So that's like, I mean, we're talking
link |
about catching the largest flying object ever made
link |
with on a giant tower with chopstick arms.
link |
It's like a karate kid with the fly, but much bigger.
link |
This probably won't work the first time.
link |
Anyway, so this is bananas.
link |
This is banana stuff.
link |
So you mentioned that you doubt, well, not you doubt,
link |
but there's days or moments when you doubt
link |
that this is even possible.
link |
It's so difficult.
link |
The possible part is, well, at this point,
link |
I think we will get Starship to work.
link |
This is a question of timing.
link |
How long will it take us to do this?
link |
How long will it take us to actually achieve
link |
full and rapid reusability?
link |
Because it will take probably many launches
link |
before we are able to have full and rapid reusability.
link |
But I can say that the physics pencils out, like we're not,
link |
at this point, I'd say we're confident that, like let's say,
link |
I'm very confident success is in the set
link |
of all possible outcomes.
link |
It's not an all set.
link |
For a while there, I was not convinced
link |
that success was in the set of possible outcomes, which
link |
is very important, actually.
link |
But so we were saying there's a chance.
link |
I'm saying there's a chance, exactly.
link |
Just not sure how long it will take.
link |
We have a very talented team.
link |
They're working night and day to make it happen.
link |
And like I said, the critical thing
link |
to achieve for the revolution in spaceflight
link |
and for humanity to be a space frame civilization
link |
is to have a fully and rapidly reusable rocket,
link |
There's not even been any orbital rocket that's
link |
been fully reusable ever.
link |
And this has always been the holy grail of rocketry.
link |
And many smart people, very smart people,
link |
have tried to do this before and have not succeeded.
link |
So because it's such a hard problem.
link |
What's your source of belief in situations like this?
link |
When the engineering problem is so difficult,
link |
there's a lot of experts, many of whom
link |
you admire who have failed in the past.
link |
And a lot of people, a lot of experts, maybe journalists,
link |
all the kind of the public in general,
link |
have a lot of doubt about whether it's possible.
link |
And you yourself know that even if it's
link |
a non null set, non empty set of success,
link |
it's still unlikely or very difficult.
link |
Like where do you go to?
link |
Both personally, intellectually as an engineer,
link |
as a team, like for source of strength,
link |
needed to sort of persevere through this
link |
and to keep going with the project,
link |
take it to completion.
link |
It's also strength.
link |
It doesn't really know how I think about things.
link |
I mean, for me, it's simply this is something
link |
that is important to get done.
link |
And we should just keep doing it or die trying.
link |
And I don't need source of strength.
link |
So quitting is not even like.
link |
That's not in my nature.
link |
And I don't care about optimism or pessimism.
link |
Fuck that, we're going to get it done.
link |
Going to get it done.
link |
Can you then zoom back in to specific problems
link |
with starship or any engineering problems you work on?
link |
Can you try to introspect your particular biological
link |
and neural network, your thinking process
link |
and describe how you think through problems,
link |
the different engineering and design problems?
link |
Is there like a systematic process you've
link |
spoken about first principles thinking,
link |
but is there a kind of process to it?
link |
Well, like saying like physics is low
link |
and everything else is a recommendation.
link |
Like I've met a lot of people that can break the law,
link |
but I haven't met anyone who could break physics.
link |
So for any kind of technology problem,
link |
you have to just make sure you're not violating physics.
link |
And first principles analysis, I think,
link |
is something that could be applied to really
link |
any walk of life, anything really.
link |
It's really just saying, let's boil something down
link |
to the most fundamental principles.
link |
The things that we are most confident
link |
are true at a foundational level.
link |
And that sets your axiomatic base.
link |
And then you reason up from there.
link |
And then you cross check your conclusion
link |
against the axiomatic truths.
link |
So some basics in physics would be like,
link |
are you violating conservation of energy or momentum
link |
or something like that?
link |
Then it's not going to work.
link |
So that's just to establish, is it possible?
link |
And another good physics tool is thinking
link |
about things in the limit.
link |
If you take a particular thing and you scale it
link |
to a very large number or to a very small number,
link |
how do things change?
link |
Well, it's like in number of things you manufacture,
link |
something like that, and then in time.
link |
Yeah, let's say you take an example of manufacturing, which
link |
I think is just a very underrated problem.
link |
And like I said, it's much harder
link |
to take an advanced technology product
link |
and bring it into volume manufacturing
link |
than it is to design it in the first place.
link |
My orders of magnitude.
link |
So let's say you're trying to figure out
link |
why is this part or product expensive?
link |
Is it because of something fundamentally foolish
link |
Or is it because our volume is too low?
link |
And then you say, OK, well, what if our volume was
link |
a million units a year?
link |
Is it still expensive?
link |
That's what I'm thinking about things in the limit.
link |
If it's still expensive at a million units a year,
link |
then volume is not the reason why your thing is expensive.
link |
There's something fundamental about design.
link |
And then you then can focus on reducing complexity
link |
or something like that in the design?
link |
Change the design to change the part to be something
link |
that is not fundamentally expensive.
link |
That's a common thing in rocketry,
link |
because the unit volume is relatively low.
link |
And so a common excuse would be, well,
link |
it's expensive because our unit volume is low.
link |
And if we were in automotive or something like that
link |
or consumer electronics, then our costs would be lower.
link |
And I'm like, OK, so let's say now you're
link |
making a million units a year.
link |
Is it still expensive?
link |
If the answer is yes, then economies of scale are not
link |
Do you throw into manufacturing?
link |
Do you throw like supply chain?
link |
Talk about resources and materials and stuff like that.
link |
Do you throw that into the calculation
link |
of trying to reason from first principles
link |
like how we're going to make the supply chain work here?
link |
And then the cost of materials, things like that.
link |
Or is that too much?
link |
Exactly, so another good example of thinking about things
link |
in the limit is if you take any product, any machine,
link |
or whatever, take a rocket or whatever
link |
and say, if you look at the raw materials in the rocket,
link |
so you're going to have aluminum, steel, titanium, incanal,
link |
specialty alloys, copper.
link |
And you say, what's the weight of the constituent elements
link |
of each of these elements?
link |
And what is their raw material value?
link |
And that sets the asymptotic limit
link |
for how low the cost of the vehicle
link |
can be unless you change the materials.
link |
And then when you do that, I call it maybe the magic one
link |
number or something like that.
link |
So that would be if you had just a pile of these raw materials
link |
here and you could wave the magic one
link |
and rearrange the atoms into the final shape,
link |
that would be the lowest possible cost
link |
that you could make this thing for
link |
unless you change the materials.
link |
So then, and that is always almost always a very low number.
link |
So then what's actually causing these to be expensive
link |
is how you put the atoms into the desired shape.
link |
Yeah, actually, if you don't mind me taking a tiny tangent,
link |
I often talk to Jim Keller, who's
link |
somebody who worked with you as a friend.
link |
Jim was a great work at Tesla.
link |
So I suppose he carries the flame of the same kind
link |
of thinking that you're talking about now.
link |
And I guess I see that same thing at Tesla and SpaceX folks
link |
who worked there, they kind of learned this way of thinking.
link |
And it kind of becomes obvious almost.
link |
But anyway, I had argument, not argument,
link |
he educated me about how cheap it
link |
might be to manufacture a Tesla bot.
link |
We just, we had an argument, how can you
link |
reduce the cost of the scale of producing a robot?
link |
Because I've gotten a chance to interact quite a bit,
link |
obviously, in the academic circles with human robots
link |
and the Boston Dynamics and stuff like that.
link |
And they're very expensive to build.
link |
And then Jim kind of schooled me on saying, OK,
link |
this kind of first principles thinking
link |
of how can we get the cost of manufacturing down?
link |
I suppose you do that, you have done that kind of thinking
link |
for Tesla bot and for all kinds of complex systems that
link |
are traditionally seen as complex.
link |
And you say, OK, how can we simplify everything now?
link |
Yeah, I mean, I think if you are really good at manufacturing,
link |
you can basically make at high volume,
link |
you can basically make anything for a cost
link |
that asymptotically approaches the raw material value
link |
of the constituents, plus any intellectual property
link |
that you need to do license.
link |
It's not like that's a very hard thing to do,
link |
but it is possible for anything.
link |
Anything in volume can be made, like I said,
link |
for a cost that asymptotically approaches
link |
its raw material constituents, plus intellectual property
link |
So what will often happen in trying to design a product
link |
is people will start with the tools and parts and methods
link |
that they are familiar with and then try
link |
to create a product using their existing tools and methods.
link |
The other way to think about it is actually
link |
imagine the try to imagine the platonic ideal of the perfect
link |
product or technology, whatever it might be.
link |
And so what is the perfect arrangement of atoms
link |
that would be the best possible product?
link |
And now let us try to figure out how
link |
to get the atoms in that shape.
link |
I mean, it sounds almost like Rick and Morty
link |
absurd until you start to really think about it.
link |
And you really should think about it in this way,
link |
because everything else is kind of,
link |
if you think you might fall victim
link |
to the momentum of the way things were done in the past,
link |
unless you think in this way.
link |
Well, just as a function of inertia,
link |
people will want to use the same tools and methods
link |
that they are familiar with.
link |
That's what they'll do by default.
link |
And then that will lead to an outcome of things
link |
that can be made with those tools and methods,
link |
that is unlikely to be the platonic ideal of the perfect
link |
So that's why it's good to think of things in both directions.
link |
They're like, what can we build with the tools that we have?
link |
But also, what is the theoretical perfect product look like?
link |
And that theoretical perfect product
link |
is going to be a moving target, because as you learn more,
link |
the definition for that perfect product will change,
link |
because you don't actually know what the perfect product is,
link |
but you can successfully approximate a more perfect
link |
So think about it like that, and then saying, OK, now,
link |
what tools, methods, materials, whatever
link |
do we need to create in order to get the atoms in that shape?
link |
But people rarely think about it that way.
link |
But it's a powerful tool.
link |
I should mention that the brilliant Siobhan Zillis is
link |
hanging out with us, in case you hear
link |
a voice of wisdom from outside, from up above.
link |
OK, so let me ask you about Mars.
link |
You mentioned it would be great for science
link |
to put a base on the moon to do some research.
link |
But the truly big leap, again, in this category
link |
of seemingly impossible, is to put a human being on Mars.
link |
When do you think SpaceX will land a human being on Mars?
link |
Best case is about five years, worst case, 10 years.
link |
What are the determining factors,
link |
would you say, from an engineering perspective,
link |
or is that not the bottlenecks?
link |
You know, it's fundamentally engineering the vehicle.
link |
I mean, Starship is the most complex and advanced rocket
link |
that's ever been made by, I don't know,
link |
water magnitude or something like that.
link |
It's really next level.
link |
So, and the fundamental optimization of Starship
link |
is minimizing cost per ton to orbit,
link |
and ultimately cost per ton to the surface of Mars.
link |
This may seem like a Mugantile objective,
link |
but it is actually the thing that needs to be optimized.
link |
Like, there is a certain cost per ton to the surface of Mars
link |
where we can afford to establish a self sustaining city.
link |
And then above that, we cannot afford to do it.
link |
So, right now, you couldn't fly to Mars for a trillion dollars.
link |
There's no amount of money could get you a ticket to Mars.
link |
So, we need to get that above, you know,
link |
to get that like something that is actually possible at all.
link |
But then, that's, we don't just want to have, you know,
link |
with Mars flags and footprints,
link |
and then not come back for a half century
link |
like we did with the moon.
link |
In order to pass a very important great filter,
link |
I think, we need to be a multi planet species.
link |
That may sound somewhat esoteric to a lot of people,
link |
but eventually, given enough time,
link |
there's something,
link |
Earth is likely to experience some calamity
link |
that could be something that humans do to themselves
link |
or an external event like happen to the dinosaurs.
link |
And eventually, if none of that happens,
link |
and somehow, magically, we keep going,
link |
then the sun will, the sun is gradually expanding
link |
and will engulf the Earth.
link |
And probably Earth gets too hot for life
link |
in about 500 million years.
link |
It's a long time, but that's only 10% longer
link |
than Earth has been around.
link |
And so, if you think about like the current situation,
link |
it's really remarkable and kind of hard to believe,
link |
but Earth's been around 4.5 billion years,
link |
and this is the first time in 4.5 billion years
link |
that it's been possible to extend life beyond Earth.
link |
And that window of charity may be open for a long time,
link |
and I hope it is, but it also may be open for a short time.
link |
And I think it was wise for us to act quickly
link |
while the window is open, just in case it closes.
link |
Yeah, the existence of nuclear weapons, pandemics,
link |
all kinds of threats should kind of give us some motivation.
link |
I mean, civilization could die with a bang or a whimper.
link |
If it dies, a demographic collapse,
link |
then it's more of a whimper, obviously.
link |
But if it's World War III, it's more of a bang.
link |
But these are all risks.
link |
I mean, it's important to think of these things
link |
and just think of things as probabilities, not certainties.
link |
There's a probability that something bad will happen on Earth.
link |
I think most likely the future will be good.
link |
But there's, let's say, for argument's sake,
link |
a 1% chance per century of a civilization ending event.
link |
Like, that was Stephen Hawking's estimate.
link |
I think he might be right about that.
link |
So then we should basically think of this
link |
like being a multiplanet species is like taking out
link |
insurance for life itself.
link |
Like, life insurance?
link |
Well, it's turned into an infomercial real quick.
link |
Life insurance for life, yes.
link |
And we can bring the creatures from plants and animals
link |
from Earth to Mars and breathe life into the planet
link |
and have a second planet with life.
link |
That would be great.
link |
They can't bring themselves there.
link |
So if we don't bring them to Mars,
link |
then they will just for sure all die when the sun expands anyway.
link |
And then that'll be it.
link |
What do you think is the most difficult aspect of building
link |
a civilization on Mars, terraforming Mars,
link |
like from an engineering perspective,
link |
from a financial perspective, human perspective,
link |
to get a large number of folks there who will never return
link |
No, they could certainly return.
link |
Some will return back to Earth.
link |
They will choose to stay there for the rest of their lives.
link |
But we need the spaceships back, like the ones that go to Mars.
link |
We need them back.
link |
So you can hop on if you want.
link |
But we can't just not have the spaceships come back.
link |
Those things are expensive.
link |
We need them back.
link |
I'd like to come back after the trip.
link |
I mean, do you think about the terraforming aspect,
link |
like actually building?
link |
Are you so focused right now on the spaceships part that's
link |
so critical to get to Mars?
link |
We absolutely, if you can't get there, nothing else matters.
link |
And like I said, we can't get there at some extraordinarily
link |
I mean, the current cost of, let's say,
link |
one ton to the surface of Mars is on the order of $1 billion.
link |
So because you don't just need the rocket and the launch
link |
and everything, you need like heat shield, you need guidance
link |
system, you need deep space communications,
link |
you need some kind of landing system.
link |
So like rough approximation would
link |
be $1 billion per ton to the surface of Mars right now.
link |
This is obviously way too expensive
link |
to create a self sustaining civilization.
link |
So we need to improve that by at least a factor of 1,000.
link |
A million per ton?
link |
Yes, ideally much less than a million ton.
link |
But if it's not, like it's got to be,
link |
obviously like how much can society afford to spend
link |
or just want to spend on a self sustaining city on Mars?
link |
The self sustaining part is important.
link |
Like it's just the key threshold,
link |
the great filter will have been passed
link |
when the city on Mars can survive even if the spaceships
link |
from Earth stop coming for any reason.
link |
It doesn't matter what the reason is.
link |
But if they stop coming for any reason,
link |
will it die out or will it not?
link |
And if there's even one critical ingredient missing,
link |
then it still doesn't count.
link |
It's like if you're in a long sea voyage
link |
and you've got everything except vitamin C,
link |
and it's only a matter of time, you know, you're going to die.
link |
So we're going to get Mars, a Mars city to the point
link |
where it's self sustaining.
link |
I'm not sure this will really happen in my lifetime,
link |
but I hope to see it at least have a lot of momentum.
link |
And then you could say, what is the minimum tonnage
link |
necessary to have a self sustaining city?
link |
And there's a lot of uncertainty about this.
link |
You could say like, I don't know,
link |
it's probably at least a million tons.
link |
Because you have to set up a lot of infrastructure on Mars.
link |
Like I said, you can't be missing anything
link |
that in order to be self sustaining, you can't be missing.
link |
Like you need a sand baking doctor, fabs,
link |
you need iron ore refineries, like you need lots of things.
link |
So, and Mars is not super hospitable.
link |
It's the least inhospitable planet,
link |
but it's definitely a fixer of a planet.
link |
Earth is pretty good.
link |
Earth is like easy.
link |
And also we should clarify in the solar system.
link |
Yes, in the solar system.
link |
There might be nice like vacation spots.
link |
There might be some great planets out there, but it's hopeless.
link |
Too hard to get there?
link |
Yeah, way, way, way, way too hard to say the least.
link |
Let me push back on that.
link |
Not really a pushback, but a quick curve ball of a question.
link |
So you did mention physics as the first starting point.
link |
So general relativity allows for warm holes.
link |
They technically can exist.
link |
Do you think those can ever be leveraged by humus
link |
to travel fast in the speed of light?
link |
Well, the one whole thing is debatable.
link |
We currently do not know of any means of going fast
link |
in the speed of light.
link |
There are some ideas about having space.
link |
So you can only move at the speed of light through space,
link |
but if you can make space itself move,
link |
that's what we're warming space.
link |
Space is capable of moving faster than the speed of light.
link |
Like the universe, in the Big Bang,
link |
the universe expanded much more than the speed of light by a lot.
link |
If this is possible, the amount of energy
link |
required to walk space is so gigantic, it boggles the mind.
link |
So all the work you've done with propulsion,
link |
how much innovation is possible with rocket propulsion?
link |
Is this, I mean, you've seen it all,
link |
and you're constantly innovating in every aspect.
link |
How much is possible?
link |
Like, how much can you get 10x somehow?
link |
Is there something in there in physics
link |
that you can get significant improvement in terms
link |
of efficiency of engines and all those kinds of things?
link |
Well, as I was saying, really the Holy Grail
link |
is a fully and rapidly reusable orbital system.
link |
So right now, the Falcon 9 is the only reusable rocket out
link |
there, but the booster comes back and lands,
link |
and you've seen the videos, and we get the nose
link |
cone fairing back, but we do not get the upper stage back.
link |
So that means that we have a minimum cost
link |
of building an upper stage.
link |
You can think of like a two stage rocket of sort
link |
of like two airplanes, like a big airplane
link |
and a small airplane, and we get the big airplane back,
link |
but not the small airplane.
link |
And so it still costs a lot.
link |
So that upper stage is at least $10 million.
link |
And then the degree of the booster is not as rapidly
link |
and completely reusable as we'd like in order of the fairings.
link |
So our kind of minimum marginal cost, not counting overhead,
link |
for per flight is on the order of $15 to $20 million maybe.
link |
So that's extremely good for, it's by far better
link |
than any rocket ever in history.
link |
But with full and rapid reusability,
link |
we can reduce the cost per ton to orbit by a factor of 100.
link |
Just think of it like imagine if you had an aircraft
link |
or something or a car.
link |
And if you had to buy in your car every time
link |
you went for a drive, it would be very expensive,
link |
every silly, frankly.
link |
But in fact, you just refuel the car or recharge the car.
link |
And that makes your trip, I don't know, 1,000 times cheaper.
link |
So it's the same for rockets.
link |
It's very difficult to make this complex machine that
link |
And so if you cannot reuse it and have
link |
to throw even any significant part of it away,
link |
that massively increases the cost.
link |
So Starship, in theory, could do a cost per launch of like
link |
a million, maybe $2 million or something like that
link |
and put over 100 tons in orbit, which is crazy.
link |
Yeah, that's incredible.
link |
So you're saying like it's by far the biggest bank for the block
link |
is to make it fully reusable versus like some kind
link |
of brilliant breakthrough in theoretical physics?
link |
There's no brilliant break.
link |
No, just make the rocket reusable.
link |
This is an extremely difficult engineering problem.
link |
But no new physics is required.
link |
Just brilliant engineering.
link |
Let me ask a slightly philosophical, fun question.
link |
I know you're focused on getting to Mars,
link |
but once we're there on Mars, what do you
link |
what form of government, economic system, political system
link |
do you think would work best for an early civilization
link |
I mean, the interesting reason to talk about this stuff,
link |
it also helps people dream about the future.
link |
I know you're really focused about the short term
link |
engineering dream, but it's like, I don't know.
link |
There's something about imagining an actual civilization
link |
on Mars that gives people, really gives people help.
link |
Well, it would be a new frontier and an opportunity
link |
to rethink the whole nature of government, just
link |
as was done in the creation of the United States.
link |
So I mean, I would suggest having direct democracy,
link |
like people vote directly on things,
link |
as opposed to representative democracy.
link |
So representative democracy, I think,
link |
is too subject to a special interest
link |
and a coercion of the politicians and that kind of thing.
link |
So I'd recommend that there's just direct democracy.
link |
People vote on laws.
link |
The population votes on laws themselves.
link |
And then the laws must be short enough
link |
that people can understand them.
link |
Yeah, and then keeping a well informed populist,
link |
really being transparent about all the information,
link |
about what they're voting for.
link |
Absolutely transparency.
link |
And not make it as annoying as those cookies
link |
where you have to accept the cookies.
link |
Like always, there's always a slight amount of trepidation
link |
when you click accept cookies.
link |
Like, I feel as though there's, perhaps,
link |
a very tiny chance that it'll open a portal to hell
link |
or something like that.
link |
That's exactly how I feel.
link |
Why do they want me to accept that?
link |
What do they want with this cookie?
link |
Like, somebody got upset with accepting cookies or something
link |
So annoying to keep accepting all these cookies.
link |
To me, this is just a grand accept.
link |
Yes, you can have my damn cookie.
link |
He heard it from me on first.
link |
He accepts all your damn cookies.
link |
It's not asking me.
link |
Yeah, it's one example of implementation
link |
of a good idea done really horribly.
link |
Yeah, it's somebody who has some good intentions
link |
of privacy or whatever.
link |
But now, everyone just has to accept cookies.
link |
And it's not, you know, you have billions of people
link |
who have to keep clicking accept cookie.
link |
It's super annoying.
link |
Then we just accept the damn cookie.
link |
There is, I think, a fundamental problem
link |
that we're, because we've not really had a major,
link |
like a world war or something like that in a while.
link |
And obviously, we'd like to not have world wars.
link |
There's not been a cleansing function
link |
for rules and regulations.
link |
So wars did have some sort of lining
link |
in that there would be a reset on rules and regulations
link |
So World War I and II, there were huge resets
link |
on rules and regulations.
link |
Now, if society does not have a war
link |
and there's no cleansing function or garbage collection
link |
for rules and regulations, then rules and regulations
link |
will accumulate every year, because they're immortal.
link |
There's no actual humans die, but the laws don't.
link |
So we need a garbage collection function
link |
for rules and regulations.
link |
They should not just be immortal,
link |
because some of the rules and regulations that are put in place
link |
will be counterproductive, done with good intentions,
link |
but counterproductive.
link |
Sometimes not done with good intentions.
link |
So if rules and regulations just accumulate every year
link |
and you get more and more of them,
link |
then eventually you won't be able to do anything.
link |
You're just like Gulliver tied down
link |
by thousands of little strings.
link |
And we see that in US and basically all economies
link |
that have been around for a while,
link |
and regulators and legislators
link |
create new rules and regulations every year,
link |
but they don't put effort into removing them.
link |
And I think that's very important that we put effort
link |
into removing rules and regulations.
link |
But it gets tough, because you get special interests
link |
that then are dependent on, like they have a vested interest
link |
in that whatever rule and regulation,
link |
and then they fight to not get it removed.
link |
Yeah, so I mean, I guess the problem with the Constitution
link |
is it's kind of like C versus Java,
link |
because it doesn't have any garbage collection built in.
link |
I think there should be,
link |
when you first said the metaphor of garbage collection,
link |
For the coding standpoint.
link |
For the coding standpoint, yeah, yeah.
link |
It would be interesting if the laws themselves
link |
kind of had a built in thing where they kind of die
link |
after a while and somebody explicitly publicly defends them.
link |
So that's sort of, it's not like somebody has to kill them,
link |
they kind of die themselves, they disappear.
link |
Not to defend Java or anything,
link |
but the C++, you could also have a great garbage collection
link |
in Python and so on.
link |
Yeah, so yeah, something needs to happen,
link |
or just the civilization's already just hardened over time.
link |
And you can just get less and less done
link |
because there's just a rule against everything.
link |
So I think like, I don't know, for Mars,
link |
whatever I say, I would say for Earth as well,
link |
like I think there should be an active process
link |
for removing rules and regulations
link |
and questioning their existence.
link |
Just like, if we've got a function
link |
for creating rules and regulations,
link |
because rules and regulations can also think of as like,
link |
they're like software or lines of code
link |
for operating civilization.
link |
That's the rules and regulations.
link |
So it's not like we shouldn't have rules and regulations,
link |
but you have code accumulation, but no code removal.
link |
And so it just gets to become basically
link |
archaic bloatware after a while.
link |
And it's just, it makes it hard for things to progress.
link |
So I don't know, maybe Mars, you'd have like any given law
link |
must have a sunset and require active voting
link |
to keep it up there, you know?
link |
And I should also say like, and these are just,
link |
I don't know, recommendations or thoughts,
link |
and ultimately we'll be up to the people in Mars
link |
to decide, but I think it should be easier
link |
to remove a law than to add one
link |
because of the, just to overcome the inertia of laws.
link |
So maybe it's like, for argument's sake,
link |
you need like say 60% vote to have a law take effect,
link |
but only a 40% vote to remove it.
link |
So let me be the guy, you posted a meme on Twitter recently
link |
where there's like a row of urinals
link |
and guy just walks all the way across
link |
and he tells you about crypto.
link |
I mean, that's happened to be so many times,
link |
I think maybe even literally.
link |
Do you think, technologically speaking,
link |
there's any room for ideas of smart contracts or so on,
link |
because you mentioned laws,
link |
that's an interesting use of things like smart contracts
link |
to implement the laws by which governments function.
link |
And like something built on Ethereum
link |
or maybe a dog coin that enables smart contracts somehow.
link |
I don't quite understand this whole smart contract thing.
link |
I mean, I'm too down to understand smart contracts.
link |
That's a good line.
link |
I mean, my general approach to any kind of like deal
link |
or whatever is just make sure there's clarity of understanding.
link |
That's the most important thing.
link |
And just keep any kind of deal very, very short
link |
and simple, plain language.
link |
And just make sure everyone understands this is the deal.
link |
Is everyone, is it clear?
link |
And what are the consequences
link |
if various things don't happen?
link |
But usually deals are business deals or whatever
link |
are way too long and complex and overly lawyered.
link |
You mentioned that Doge is the people's coin.
link |
And you said that you were literally going SpaceX,
link |
may consider literally putting a Doge coin on the moon.
link |
Is this something you're still considering Mars perhaps?
link |
Do you think there's some chance
link |
we've talked about political systems on Mars
link |
that Doge coin is the official currency of Mars
link |
at some point in the future?
link |
Well, I think Mars itself will need to have
link |
a different currency because you can't synchronize
link |
due to speed of light or not easily.
link |
So it must be completely standalone from Earth?
link |
Well, yeah, because Mars is at closest approach,
link |
it's four light minutes away roughly.
link |
And then at most approach,
link |
it's roughly 20 light minutes away, maybe a little more.
link |
So you can't really have something synchronizing.
link |
You know, if you've got a 20 minutes speed of light issue,
link |
if it's got a one minute block chain,
link |
it's not gonna synchronize probably.
link |
So Mars, I don't know if Mars would have a crypto currency
link |
as a thing, but probably seems likely,
link |
but it would be some kind of localized thing on Mars.
link |
And you let the people decide?
link |
The future of Mars should be up to the Martians.
link |
Yeah, so I mean, I think the crypto currency thing
link |
is an interesting approach to reducing the error
link |
in the database that is called money.
link |
You know, I think I have a pretty deep understanding
link |
of what money actually is on a practical day to day basis
link |
because of PayPal.
link |
You know, we really got in deep there.
link |
And right now the money system,
link |
actually for practical purposes,
link |
is really a bunch of heterogeneous mainframes
link |
running old cobalt.
link |
Okay, you mean literally?
link |
Literally what's happening.
link |
Okay, in batch mode.
link |
Yeah, pretty the poor bastards
link |
who have to maintain that code.
link |
Okay, that's a pain.
link |
Not even Fortran, it's cobalt, yep.
link |
And the banks are still buying mainframes in 2021
link |
and running ancient cobalt code.
link |
And the Federal Reserve is probably even older
link |
than what the banks have
link |
and they have an old cobalt mainframe.
link |
And so now, and so the government effectively
link |
has editing privileges on the money database.
link |
And they use those editing privileges
link |
to make more money whenever they want.
link |
And this increases the error in the database that is money.
link |
So I think money should really be viewed
link |
through the lens of information theory.
link |
And so it's, you're kind of like an internet connection.
link |
Like what's the bandwidth, you know, total bit rate?
link |
What is the latency, jitter, packet drop,
link |
you know, errors in network communication?
link |
Just think of money like that, basically.
link |
I think that's probably why we really think of it.
link |
And then say what system
link |
from an information theory standpoint
link |
allows an economy to function the best.
link |
And, you know, crypto is an attempt to reduce the error
link |
in money that is contributed by governments
link |
diluting the money supply
link |
as basically a pernicious form of taxation.
link |
So both policy in terms of with inflation and actual,
link |
like technological cobalt, like cryptocurrency
link |
takes us into the 21st century
link |
in terms of the actual systems
link |
that allow you to do the transaction,
link |
to store wealth, all those kinds of things.
link |
Like I said, just think of money as information.
link |
People often will think of money
link |
as having power in and of itself.
link |
Money is information.
link |
And it does not have power in and of itself.
link |
Like, you know, applying the physics tools
link |
of thinking about things in the limit is helpful.
link |
If you are stranded on a tropical island
link |
and you have a trillion dollars, it's useless.
link |
Because there's no resource allocation.
link |
Money is a database for resource allocation.
link |
If there's no resource to allocate except yourself.
link |
So money is useless.
link |
If you're stranded on a desert island with no food,
link |
all the Bitcoin in the world will not stop you from starving.
link |
So just think of money as a database
link |
for resource allocation across time and space.
link |
And then what system in what form should that database
link |
or data system, what would be most effective?
link |
Now, there is a fundamental issue with, say, Bitcoin
link |
in its current form in that the transaction volume
link |
And the latency for a properly confirmed transaction
link |
is too long, much longer than you'd like.
link |
So it's actually not great from a transaction volume
link |
standpoint or a latency standpoint.
link |
So it is perhaps useful to solve an aspect
link |
of the money database problem,
link |
which is the sort of store of wealth
link |
or an accounting of relative obligations, I suppose.
link |
But it is not useful as a currency,
link |
as a day to day currency.
link |
But people have proposed different technological solutions.
link |
Yeah, Lightning Network and the layer two technologies
link |
I mean, it's all, it seems to be all kind of a trade off.
link |
But the point is, it's kind of brilliant to say
link |
that just think about it information,
link |
think about what kind of database,
link |
what kind of infrastructure enables
link |
the exchange of information.
link |
Like if you're operating in an economy,
link |
and you need to have some thing that allows
link |
for the efficient, to have efficient value ratios
link |
between products and services.
link |
So you've got this massive number of products
link |
and services and you need to, you can't just barter.
link |
It's like, that would be extremely unwieldy.
link |
So you need something that gives you a ratio of exchange
link |
between goods and services.
link |
And then something that allows you to shift obligations
link |
across time, like debt, debt and equity,
link |
shift obligations across time.
link |
Then what does the best job of that?
link |
Part of the reason why I think there's some
link |
marriage to dogecoin, even though it was obviously created
link |
as a joke, is that it actually does have
link |
a much higher transaction volume capability than Bitcoin.
link |
And the costs of doing a transaction,
link |
the dogecoin fee is very low.
link |
Like right now, if you want to do a Bitcoin transaction,
link |
the price of doing that transaction is very high.
link |
So you could not use it effectively for most things.
link |
And nor could it even scale to a high volume.
link |
And when Bitcoin started, I guess around 2008
link |
or something like that, the internet connections
link |
were much worse than they are today.
link |
Like order of magnitude, I mean, there's the way,
link |
way worse in 2008.
link |
So like having a small block size or whatever
link |
is, and a long synchronization time
link |
is made sense in 2008.
link |
But to 2021 or fast forward 10 years,
link |
it's like, it's like comically low, it's a,
link |
so, and I think there's some value
link |
to having a linear increase in the amount of currency
link |
that is generated.
link |
So because some amount of the currency,
link |
like if a currency is too deflationary or like,
link |
or should say if a currency is expected
link |
to increase in value over time,
link |
there's reluctance to spend it.
link |
Cause you're like, oh, if I,
link |
I'll just hold it and not spend it
link |
because it's scarcity is increasing with time.
link |
So if I spend it now, then I will regret spending it.
link |
So I will just, you know, total it.
link |
But if there's some dilution of the currency
link |
occurring over time, that's more of an incentive
link |
to use it as a currency.
link |
So those coins somewhat randomly has a,
link |
just a fixed number of sort of coins
link |
or hash strings that are generated every year.
link |
So there's some inflation, but it's not a percentage base.
link |
It's a fixed number.
link |
So the percentage of inflation
link |
will necessarily decline over time.
link |
So it just, I'm not saying
link |
that it's like the ideal system for a currency,
link |
but I think it actually is just fundamentally better
link |
than anything else I've seen, just by accident.
link |
Like I said, around 2008.
link |
So you're not, you know, some people suggested
link |
you might be set to Oshinakamoto.
link |
You've previously said you're not, let me ask.
link |
You're not for sure.
link |
Would you tell us if you were?
link |
Do you think it's a feature or bug
link |
that he's anonymous or she or they?
link |
It's an interesting kind of quirk of human history
link |
that there is a particular technology
link |
that is a completely anonymous inventor.
link |
Well, I mean, you can look at the evolution of ideas
link |
before the launch of Bitcoin
link |
and see who wrote, you know, about those ideas.
link |
And then I don't know exactly,
link |
obviously I don't know who created Bitcoin
link |
for practical purposes,
link |
but the evolution of ideas is pretty clear for that.
link |
And like it seems as though like Nick Szabo
link |
is probably more than anyone else
link |
responsible for the evolution of those ideas.
link |
So he claims not to be Nakamoto,
link |
but I'm not sure that's neither here nor there,
link |
but he seems to be the one more responsible
link |
for the ideas behind Bitcoin than anyone else.
link |
So it's not perhaps like singular figures
link |
aren't even as important as the figures involved
link |
in the evolution of ideas that led to a thing.
link |
So, you know, most perhaps it's sad to think about history,
link |
but maybe most names will be forgotten anyway.
link |
What is the name anyway?
link |
It's a name attached to an idea.
link |
What does it even mean really?
link |
I think Shakespeare had a thing about roses and stuff,
link |
whatever you said.
link |
Roses by any other name.
link |
I gotta yield on to quote Shakespeare.
link |
I feel like I accomplished something today.
link |
Shall I compare thee to a summer's day?
link |
I'm gonna clip that out instead of doing it.
link |
It's a lot more temperate and more fair.
link |
Autopilot, Tesla autopilot.
link |
Tesla autopilot has been through an incredible journey
link |
over the past six years,
link |
or perhaps even longer in the minds of,
link |
in your mind, in the minds of many involved.
link |
I think that's where we first like connected
link |
really was the autopilot stuff, autonomy and.
link |
The whole journey was incredible to me to watch.
link |
because I knew, well, part of it was I was at MIT
link |
and I knew the difficulty of computer vision.
link |
And I knew the whole, I had a lot of colleagues
link |
and friends about the DARPA challenge.
link |
I knew how difficult it is.
link |
And so there was a natural skepticism.
link |
When I first drove a Tesla with the initial system
link |
based on Mobileye, I thought there's no way.
link |
So the first one I got in, I thought there's no way
link |
this car could maintain, like staying in the lane
link |
and create a comfortable experience.
link |
So my intuition initially was that the lane keeping problem
link |
is way too difficult to solve.
link |
Oh, lane keeping, yeah, that's relatively easy.
link |
Well, like, but not the, but solve in the way that we just,
link |
we talked about previous is prototype versus a thing
link |
that actually creates a pleasant experience
link |
over hundreds of thousands of miles and millions.
link |
I mean, we had to wrap a lot of code around the Mobileye thing.
link |
It doesn't just work by itself.
link |
I mean, that's part of the story
link |
of how you approach things sometimes.
link |
Sometimes you do things from scratch.
link |
Sometimes at first you kind of see what's out there
link |
and then you decide to do from scratch.
link |
That was one of the boldest decisions I've seen
link |
is both on the hardware and the software
link |
to decide to eventually go from scratch.
link |
I thought, again, I was skeptical
link |
of whether that's going to be able to work out
link |
because it's such a difficult problem.
link |
And so it was an incredible journey.
link |
What I see now with everything,
link |
the hardware, the compute, the sensors,
link |
the things I maybe care and love about most
link |
is the stuff that Andre Carpathi is leading
link |
with the data set selection,
link |
the whole data engine process,
link |
the neural network architectures,
link |
the way that's in the real world,
link |
that network is tested, validated,
link |
all the different test sets,
link |
versus the ImageNet model of computer vision
link |
like what's in academia is like real world
link |
artificial intelligence.
link |
So, and Andre is awesome
link |
and obviously plays an important role,
link |
but we have a lot of really talented people driving things.
link |
So, and Ashok is actually the head
link |
of autopilot engineering.
link |
Andre is the director of AI.
link |
AI stuff, yeah, yeah.
link |
So yeah, there's, I'm aware that there's an incredible team
link |
of just a lot going on.
link |
Yeah, I just, you know, people will give me too much credit
link |
and they'll give Andre too much credit, so.
link |
And people should realize how much is going on
link |
Yeah, it's just a lot of really talented people.
link |
The Tesla autopilot AI team is extremely talented.
link |
It's like some of the smartest people in the world.
link |
So yeah, we're getting it done.
link |
What are some insights you've gained
link |
over those five, six years of autopilot
link |
about the problem of autonomous driving?
link |
So, you leaped in having some sort of first principles,
link |
kinds of intuitions, but nobody knows
link |
how difficult the problem, like the problem.
link |
I thought the self driving problem would be hard,
link |
but it was harder than I thought.
link |
It's not like I thought it'd be easy.
link |
I thought it'd be very hard,
link |
but it was actually way harder than even that.
link |
So, what it comes down to at the end of the day
link |
is to solve self driving, you have to solve,
link |
you basically need to recreate what humans do to drive,
link |
which is humans drive with optical sensors,
link |
eyes, and biological neural nets.
link |
And so in order to, that's how the entire road system
link |
is designed to work with basically passive optical
link |
and neural nets, biologically.
link |
And now that we need to,
link |
so for actually for full self driving to work,
link |
we have to recreate that in digital form.
link |
So we have to, that means cameras
link |
with advanced neural nets in silicon form,
link |
and then it will obviously solve for full self driving.
link |
That's the only way.
link |
I don't think there's any other way.
link |
But the question is, what aspects of human nature
link |
do you have to encode into the machine, right?
link |
Do you have to solve the perception problem, like detect?
link |
And then you first, while it realize,
link |
what is the perception problem for driving?
link |
Like all the kinds of things you have to be able to see.
link |
Like what do we even look at when we drive?
link |
There's, I just recently heard Andre talked about at MIT
link |
about like car doors.
link |
I think it was the world's greatest talk
link |
of all time about car doors.
link |
The fine details of car doors.
link |
Like what is even an open car door, man?
link |
So like the the ontology of that,
link |
that's the perception problem.
link |
We humans solve that perception problem
link |
and Tesla has to solve that problem.
link |
And then there's the control and the planning
link |
coupled with the perception.
link |
You have to figure out like what's involved in driving,
link |
like especially in all the different edge cases.
link |
And then, I mean, maybe you can comment on this,
link |
how much game theoretic kind of stuff needs to be involved,
link |
you know, at a four way stop sign.
link |
You know, as humans, when we drive,
link |
our actions affect the world.
link |
Like it changes how others behave.
link |
Most of the time was driving,
link |
if you're usually just responding to the scene,
link |
as opposed to like really asserting yourself in the scene.
link |
I think these sort of control logic conundrums
link |
are not the hard part.
link |
The, you know, let's see.
link |
What do you think is the hard part
link |
in this whole beautiful, complex problem?
link |
So it's a lot of frigging software, man.
link |
A lot of smart lines of code.
link |
For sure, in order to have create an accurate vector space.
link |
So like you're coming from image space,
link |
which is like this flow of photons.
link |
You're going to the camera cameras
link |
and then you have this massive bit stream in image space.
link |
And then you have to effectively compress
link |
a massive bit stream corresponding to photons
link |
that knocked off an electron in a camera sensor
link |
and turn that bit stream into vector space.
link |
By vector space, I mean like,
link |
you know, you've got cars and humans
link |
and lane lines and curves and traffic lights
link |
and that kind of thing.
link |
Once you have an accurate vector space,
link |
the control problem is similar to that of a video game,
link |
like a grand theft order of cyberpunk.
link |
If you have accurate, accurate, best vector space.
link |
It's the control problem is,
link |
I wouldn't say it's trivial, it's not trivial, but it's,
link |
it's not like some insurmountable thing.
link |
But having an accurate vector space is very difficult.
link |
Yeah, I think we humans don't give enough respect
link |
to how incredibly human perception system is,
link |
to mapping the raw photons
link |
to the vector space representation in our heads.
link |
Your brain is doing an incredible amount of processing
link |
and giving you an image that is a very cleaned up image.
link |
Like when we look around here, we see,
link |
like you see color in the corners of your eyes,
link |
but actually your eyes have very few cones,
link |
like cone receptors in the peripheral vision.
link |
Your eyes are painting color in the peripheral vision.
link |
You don't realize it, but their eyes
link |
are actually painting color.
link |
And your eyes also have like this blood vessels
link |
and also to gnarly things.
link |
And there's a blind spot, but do you see your blind spot?
link |
No, your brain is painting in the missing, the blind spot.
link |
You can do these like, see these things online
link |
where you look here and look at this point
link |
and then look at this point.
link |
And it's, if it's in your blind spot,
link |
your brain will just fill in the missing bits.
link |
It's so cool. The peripheral vision is so cool.
link |
It makes you realize all the illusions for vision science
link |
and so it makes you realize just how incredible the brain is.
link |
The brain is doing crazy amount of post processing
link |
on the vision signals from your eyes.
link |
So, and then even once you get all those vision signals,
link |
your brain is constantly trying to forget as much as possible.
link |
So, human memory is perhaps the weakest thing
link |
about the brain is memory.
link |
So, because memory is so expensive to a brain
link |
and so limited, your brain is trying to forget
link |
as much as possible and distill the things that you see
link |
into the smallest amounts of information possible.
link |
So, your brain is trying to not just get to a vector space,
link |
but get to a vector space
link |
that is the smallest possible vector space
link |
of only relevant objects.
link |
And I think like, you can sort of look inside your brain,
link |
or at least I can, like when you drive down the road
link |
and try to think about what your brain is actually doing
link |
And it's, it's, it's, it's, it's, it's like, you'll see a car
link |
that's, because you're, you're, you don't have cameras.
link |
You, I don't have eyes in the back of your head or the side.
link |
You know, so you say like, you basically, your, your head is
link |
like a, you know, you basically have like two cameras
link |
And, and what's your, and I said, it's not that great.
link |
You and I is, you know, like, and people are constantly
link |
distracted and thinking about things and texting
link |
and doing all sorts of things they shouldn't do in a car,
link |
changing the radio station.
link |
So, having arguments, you know, is like, so, so then,
link |
like, say like, like, like when's the last time you looked
link |
right and left and, you know, or, and, and rearward or even
link |
diagonally, you know, forward to actually refresh your vector
link |
So you're glancing around and what your mind is doing is, is,
link |
is trying to still the relevant vectors, basically objects
link |
with a position and motion and, and, and then, and then, and
link |
then editing that down to the least amount that's necessary
link |
It does seem to be able to edit it down or compress it even
link |
further into things like concepts.
link |
So it's not, it's like, it goes beyond the human mind seems
link |
to go sometimes beyond vector space to sort of space of
link |
concepts to where you'll see a thing.
link |
It's no longer represented spatially somehow.
link |
It's almost like a concept that you should be aware of.
link |
Like if this is a school zone, you'll remember that.
link |
As a concept, which is a weird thing to represent, but
link |
perhaps for driving, you don't need to fully represent
link |
those things, or maybe you get those kind of, um,
link |
well, you, you, you, you, you need to like establish vector
link |
space and then actually have predictions for, uh, that those
link |
So like, um, you know, like if, uh, you know, like you drive
link |
fast, say, say, uh, uh, uh, uh, uh, a bus and the, and you
link |
see that this, this people, before you drove past the bus,
link |
you saw people crossing like, or some just imagine there's
link |
like a large truck or something blocking site.
link |
Um, but you, before you came out to the truck, you saw
link |
that there were some kids about to cross the road in front
link |
Now you can no longer see the kids, but you, you, you need
link |
to be able, but you would now know, okay, those kids are
link |
probably going to pass by the truck and cross the road, even
link |
though you cannot see them.
link |
So you have to have, um, memory, uh, you have to need to
link |
remember that there were kids there and you need to have
link |
some forward prediction of what their position will be.
link |
It's a really hard problem at the time of relevance.
link |
So with, with occlusions and computer vision, when you can't
link |
see an object anymore, even when it just walks behind a tree
link |
and reappears, that's a really, really, I mean, at least in
link |
academic literature, it's tracking through occlusions.
link |
It's very difficult.
link |
Um, I understand this.
link |
So some of it, it's like object permanence, like the same
link |
thing happens with humans, with neural nets, like when, like
link |
a toddler grows up, like there's a, there's a point in time
link |
where, uh, they develop, they have a sense of object
link |
So before a certain age, if you have a ball, uh, or a toy
link |
or whatever, and you put it behind your back and you pop it
link |
out, if they don't, before they have object permanence, it's
link |
like a new thing every time.
link |
It's like, whoa, this toy went, poof, just spared.
link |
And now it's back again.
link |
And they can't believe it.
link |
And that they can play peekaboo all day long because the
link |
peekaboo is fresh every time.
link |
But then we figured out object permanence.
link |
Then they realized, oh, no, the object is not gone.
link |
It's just behind your back.
link |
Um, sometimes I wish we never did figure out object permanence.
link |
So that's, uh, that's an important problem to solve.
link |
So, so, and like an important evolution of the neural nets in
link |
the car is, uh, um, memory across, memory across both time and
link |
Um, so, uh, no, you can't remember, like you have to say,
link |
like how long do you want to remember things for?
link |
And, and it's, it doesn't, there's a cost to remembering
link |
things for a long time.
link |
So you get, you know, like run out of memory to try to remember
link |
too much for too long.
link |
Um, and, and then you also have things that are stale.
link |
If, if, if they're, if you remember them for too long, and
link |
then you also need things that are remembered, uh, remembered
link |
So even if you like, say, have like, for a good sake, five
link |
seconds of memory, uh, on a time basis, but like, let's say
link |
you, you, you're parked at a light and you, and you saw, you
link |
use a pedestrian example that people were waiting to cross
link |
the, across the road and you can't, you can't quite see them
link |
because of an occlusion.
link |
Uh, but they might wait for a minute before the light
link |
changes for them to cross the road.
link |
You still need to, to remember that they, that that's where
link |
they were, um, and that they're probably going to cross the
link |
road type of thing.
link |
Um, so even if that exceeds your, your, your time based
link |
memory should not exceed your space of memory.
link |
And I just think the data engine side of that.
link |
So getting the data to learn all of the concepts that you're
link |
saying now is an incredible process.
link |
It's this iterative process of just, it's this, this
link |
hydranet of many, hydranets, we're changing the name to
link |
I'm sure it'd be equally as Rick and Morty like, yeah, we've
link |
rearchitected the neural net, uh, neural nets in the cars.
link |
So many times it's crazy.
link |
Oh, so every time there's a new major version, you'll rename it
link |
to something more ridiculous or, or memorable and beautiful.
link |
Sorry, not ridiculous, of course.
link |
If you, if you see the full, the full like, uh, array of neural
link |
nets that, that, that are operating in the car, it's kind
link |
of boggles the mind.
link |
There's so, there's so many layers.
link |
Um, but, and, and we, we started off with, uh, simple neural
link |
nets that were, uh, basically image recognition on a single
link |
frame, from a single camera, uh, and then, uh, trying to knit
link |
those together with it, you know, it with a C, uh, I should
link |
say we, we're really primarily running C here because C plus
link |
plus is too much overhead and we have our own C compiler.
link |
So to get maximum performance, we actually wrote our own C
link |
compiler and are continuing to optimize our C compiler, uh, for
link |
maximum efficiency.
link |
In fact, we've just recently, uh, done a new river on a, on a
link |
C compiler that will compile directly to our autopilot
link |
Do you want to compile the whole thing down and with your
link |
Like so efficiency here, cause there's all kinds of compute.
link |
There's CPU, GPU, there's like the ASIC type of thing that's,
link |
and you have to somehow figure out the scheduling across all
link |
And so you're compiling the code down.
link |
It does all the, okay.
link |
This is, so that's why there's a lot of people involved.
link |
There's, there's a lot of hardcore, uh, software engineering at a
link |
very sort of bare metal level, uh, cause you, we're trying to do
link |
a lot of compute, uh, that's constrained to the, you know, our
link |
full self driving computer.
link |
So, and we want to try to have the highest frames per second, um,
link |
possible, um, with, with sort of very, very finite amount of
link |
compute, um, and power.
link |
So, um, we really put a lot of effort into the efficiency
link |
Um, and, and, uh, so there's actually a lot of work done by some
link |
very talented software engineers at Tesla that, uh, at a very
link |
foundational level to improve the efficiency of compute and how
link |
we use the, the, the trip accelerators, uh, which are basically,
link |
um, dot, you know, uh, doing matrix math dot, dot products, uh,
link |
like a bazillion dot products.
link |
And it's like, what, what, what, what are neural nets?
link |
It's like computer wise, like 99% dot products.
link |
And you want to achieve as many high frame rates like a video game.
link |
You want full resolution, high frame rate, high frame rate, low latency,
link |
um, low jitter, uh, so, um, I think one of the things for, um, moving
link |
towards now is no post processing of the image through the, um, uh, the
link |
image signal processor.
link |
So, um, like for, for what happens for cameras is that almost all
link |
cameras is they, um, there's a lot of post processing done in order
link |
to make pictures look pretty.
link |
Uh, and so we don't care about pictures looking pretty.
link |
Um, we, we just want the data.
link |
We, we, so we're, we're moving to just roll, roll photon counts.
link |
So the system will, like the image that, that, that the computer sees is
link |
actually much more than what you'd see if you're represented on a camera.
link |
It's got much more data.
link |
Uh, and even in very low light conditions, you can see that there's
link |
a small photon count difference between, you know, the spot here and
link |
that's about there, which means that, so it can see in the dark incredibly
link |
well, um, because it can detect these tiny differences in photon counts.
link |
Much better than you'd possibly imagine.
link |
Um, so, and then we also save, uh, 13 milliseconds on a latency.
link |
Uh, so, uh, from removing the post processing and the image.
link |
It's like, um, because we've got eight cameras and, and then there's, uh,
link |
roughly, I don't know, one and a half milliseconds or so, maybe 1.6 milliseconds
link |
of latency, um, for each camera.
link |
And so it, like, um, going to just, uh, it basically bypassing the image processor.
link |
Uh, gets us back 13 milliseconds of latency, which is important, um, and we
link |
track latency all the way from, you know, photon hits the, the camera to, you
link |
know, all the steps that it's got to go through to get, you know, go through the,
link |
um, the various neural nets and the, the C code and, uh, and there's a
link |
little bit of C plus plus there as well.
link |
Um, well, I can maybe a lot, but it, the core stuff is the heavy duty computers
link |
all in C, um, and, uh, and so, so we track that latency all the way to an
link |
output command to the, um, drive unit to accelerate, uh, the brakes just to slow
link |
down, steering your turn left or right.
link |
Um, so, cause you got to output a command that's going to go to a controller
link |
and like some of these controllers have an update frequency that's maybe, uh, 10
link |
hertz or something like that, which is slow.
link |
That's like, now you lose a hundred milliseconds potentially.
link |
So, um, so then we want to update the, the drivers on the, like, say, steering and
link |
braking control to have, um, more like, uh, 100 hertz instead of 10 hertz and you
link |
got a 10 millisecond latency instead of a hundred milliseconds worst case latency.
link |
And actually jitter is more of a challenge than, than, than latency.
link |
Because latency is like, you can, you can, you can anticipate and predict, but if
link |
you're, but if you've got a stack up of things going from the camera to the, to
link |
the computer through, then you can, you can, you can, you can anticipate and
link |
the computer through then a series of other computers.
link |
And finally to an actuator on the car.
link |
If you have a stack up of, uh, of tolerances of timing tolerances, then you
link |
can have quite a variable latency, which is called jitter.
link |
And, and that makes it a hard to, to, to anticipate exactly what, how you should
link |
turn the car or accelerate.
link |
Because, you know, if you've got maybe 150 to 200 milliseconds of jitter, then
link |
you could be off by, you know, up to.2 seconds.
link |
And this could make, this could make a big difference.
link |
So you have to interpolate somehow to, to, to, to, uh, deal with the effects of jitter.
link |
So you, you, you can make like robust control decisions.
link |
Again, you have to, uh, so the jitter is in the sensor information or is it, the jitter
link |
can occur at any stage in the pipeline.
link |
You can, if you have just, if you have a fixed latency, you can anticipate, um, and,
link |
and, and, uh, like say, okay, we know that, uh, our information is, for argument's sake,
link |
150 milliseconds stale.
link |
Like, so for, for, um, 140, for argument's sake, 150 milliseconds from photon second
link |
camera to, um, where you can measure a change in the acceleration of the vehicle.
link |
So then, uh, then you're going to say, okay, well, we're going to enter, we know it's
link |
150 milliseconds, so we're going to take that into account and, uh, and compensate for that
link |
However, if you, if you've got then 150 milliseconds of latency plus 100 milliseconds of jitter,
link |
that's, which could be anywhere from zero to 100 milliseconds on top.
link |
So, so then your latency could be from 150 to 250 milliseconds.
link |
Now you've got 100 milliseconds that you don't know what to do with and, and that's basically random.
link |
So getting rid of jitter is extremely important.
link |
And that affects your control decisions and all those kinds of things.
link |
Okay. Yeah, the, the cars is going to fundamentally maneuver better with lower jitter.
link |
And the, the, the, the cars will maneuver with superhuman ability and reaction time
link |
much faster than a human.
link |
I mean, I think over time the autopilot, full cell driving will be capable of maneuvers that,
link |
you know, are far more than what like James Bond could do in like the best movie type of thing.
link |
That's exactly where I was imagining my mind as you said it.
link |
It's like impossible maneuvers that a human couldn't do.
link |
Well, let me ask sort of, uh, looking back the six years, looking out into the future,
link |
based on your current understanding, how, how hard do you think this,
link |
this full self driving problem, when do you think Tesla will solve level four FSD?
link |
I mean, it's looking quite likely that it will be next year.
link |
And what does the solution look like? Is it the current pool of FSD beta candidates?
link |
They start getting greater and greater as they have been degrees of autonomy.
link |
And then there's a certain level beyond which they can, they can do their own, they can read a book.
link |
Yeah. So, uh, I mean, you can see that anybody who's been following the
link |
full self driving beta closely will see that the, um, the rate of disengagement has been
link |
dropping rapidly. So like disengagement be where, where the driver intervenes to prevent the car
link |
from doing something dangerous potentially. So, um, so the interventions, you know, per million
link |
miles has been dropping dramatically at some point the, and that trend looks like it happens next
link |
year is the, the, the, the probability of an accident on FSD is less than that of the average
link |
human and then, and then significantly less than that of the average human. Um, so it certainly
link |
appears like we will get there next year. Um, then, then of course that, that, then there's
link |
going to be a case of, okay, well, we now have to prove this to regulators and prove it to,
link |
you know, and, and we, we, we want a standard that is not just equivalent to a human, but
link |
uh, much better than the average human. I think it's got to be at least two or three times, uh,
link |
higher safety than a human. So two or three times lower probability of injury than a human.
link |
Um, before, before we would actually say like, okay, it's okay to go. It's not going to be a
link |
cool, it's going to be much better. So if you look at 10 point FSD, 10.6 just came out recently,
link |
10.7 is on the way, maybe 11 is on the way to where in the future. Yeah. Um, we were hoping
link |
to get 11 out this year, but it's, uh, 11 actually has a whole bunch of, uh, fundamental
link |
rewrites on the neural, neural net architecture, um, and, and some fundamental improvements, uh, in
link |
creating vector space. Uh, so, uh, there is a, some fundamental like leap that really deserves
link |
the 11. I mean, that's a pretty cool number. Yeah. Yeah. Uh, 11 would be a single stack for
link |
all, you know, one stack to rule them all. Um, and, uh, but, but there, there's just some really
link |
fundamental, uh, neural net architecture changes that are, that will allow for, uh, much more
link |
capability, but, but, you know, at first they're going to have issues. So like we have this working
link |
on like sort of alpha software and it's good, but it's, uh, it's, it's, it's, it's basically taking
link |
a whole bunch of C C plus plus code and, and, and leading a massive amount of C plus plus code
link |
and replacing it with the neural net. And you know, Andre, um, makes this point a lot, which
link |
he's like neural nets, that kind of eating software, you know, over time there's like
link |
less and less conventional software, more and more neural net, uh, which is still software, but it's,
link |
you know, still comes out the lines of software, but, uh, it's more, more neural net stuff, uh, and
link |
less, uh, you know, heuristics basically, um, if you're more, more, more, uh, matrix based
link |
stuff, unless, uh, heuristics based stuff. Um, and, um, you know, like, like, like one of the big changes
link |
will be, um, like right now the neural nets, uh, will, um, deliver a giant bag of points
link |
to the C plus plus or C and C plus plus code. Yeah. Um, we call it the giant bag of points.
link |
Yeah. Uh, and it's like, so you go to pixel and, and, and, and something associated with that pixel,
link |
like this pixel is probably car, the pixel is probably lane line. Um, then you've got to
link |
assemble this giant bag of points in the C code and turn it into, uh, vectors. Um, and, uh,
link |
it does a pretty good job of it, but it's, it's, uh, it's, we want to just, we need another layer
link |
of neural nets on top of that to take the, the giant bag of points and distill that down to
link |
vector space in the, in the neural net part of the software as opposed to the heuristics
link |
part of the software. This is a big improvement. Um, neural nets all the way down. That's what
link |
you want. It's not even all neural nets, but it's, it's, it's, uh, this will be just a, this
link |
is a game changer to not have the bag of points, the giant bag of points that has to be assembled
link |
with, um, many lines of C C plus plus, uh, and, and have the, and have a neural net just
link |
assemble those into vectors. So, so the, the neural net is outputting, um, much, much less
link |
data. It's, it's, it's outputting this, this is a lane line. This is a curb. This is drivable
link |
space. This is a car. This is, uh, you know, a pedestrian or cyclist or something like that.
link |
It's outputting, um, it's really outputting, um, proper vectors to the, the C C plus plus control
link |
control code as opposed to the sort of constructing the, the vectors, uh, in, in C. Um, we're done,
link |
I think, quite a good job of, but it's, it's a, it's kind of hitting a local maximum on the,
link |
how well the C can do this. Um, so this is, this is really, this is really a big deal. And, and
link |
just all of the networks in the car need, need to move to surround video. There's still some
link |
legacy networks that are not, uh, surround video. Um, and all of the training needs to move to
link |
surround video and the efficiency of the training, uh, it needs to get better than it is. Uh, and
link |
then we need to move everything to, uh, raw, uh, photon, uh, counts as opposed to, um, processed
link |
images. Okay. It's just, it's just quite a big reset on the training because the system's trained
link |
on post processed image images. So we need to redo all the training, uh, to train against
link |
the, the raw photon counts instead of the post processed image. So ultimately it's kind of
link |
reducing the complexity of the whole thing. So, uh, reducing, reducing the lines of code will
link |
actually go, go lower. Yeah. That's fascinating. Um, so you're doing fusion of all the sensors
link |
and reducing the complexity of having to deal with these cameras. There's a lot of cameras
link |
really. Right. Yes. Um, same with humans. Uh, well, I guess we got years too. Okay. Yeah.
link |
Well, we'll actually need to incorporate, um, sound as well. Um, cause you know, you need to like
link |
listen for ambulance sirens or fire, you know, fire trucks, you know, uh, if somebody like
link |
you know, yelling at you or something, I don't know, just that there's, there's a little bit of
link |
audio that needs to be incorporated as well. Do you need to go back to break? Yeah, let's
link |
just, let's take a break. Okay. Honestly, frankly, like the ideas are, are the easy thing and the
link |
implementation is the hard thing. Like the idea of going to the moon is, is the easy part,
link |
but going to the moon is the hard part. It's the hard part. Um, and there's a lot of like hardcore
link |
engineering that's got to get done at the hardware and software level. Uh, likes optimizing the
link |
C compiler and, uh, just, you know, uh, cutting out latency everywhere. Like this is, if we don't
link |
do this, the system will not work properly. Um, so the work of the engineers doing this,
link |
they are like the unsung heroes to some, you know, but they are critical to the success of the
link |
situation. I think he made it clear. I mean, at least to me, it's super exciting. Everything
link |
that's going on outside of what Andre is doing. Yeah. Just the whole infrastructure, the software.
link |
I mean, everything is going on with data engine, uh, whatever, whatever it's called,
link |
the whole process is, is just work of art to me.
link |
Yeah. I think the, the, the sure scale of it is, is boggles mind. Like the training,
link |
the amount of work done with the, like we've written all this custom software for training
link |
and labeling, um, and to do auto labeling, auto labeling is essential. Um, because especially
link |
when you've got like surround video, it's very difficult to like label surround video from scratch
link |
is extremely difficult. Um, like take a human's such a long time to even label one video clip
link |
like several hours, uh, or the order label it, uh, basically we're just apply a heavy, like heavy duty,
link |
uh, like a lot of compute to the, to the video clips, um, to pre assign and guess what all the
link |
things are that are going on in this round video. And then there's like correcting it.
link |
Yeah. And then all the human has to do is like tweet, like say the, you know, change, adjust
link |
what is incorrect. This, this is like increase, increase this productivity by effect a hundred
link |
or more. Yeah. Uh, so you've presented Tesla bot as primarily useful in the factory. First of all,
link |
I think human robots are incredible from a fan of robotics. I think, uh, the elegance of movement
link |
that human, um, the human robots that by Peter robots show are just so cool. So it's, uh, really
link |
interesting that you're working on this and also talking about applying the same kind of all the
link |
ideas of some of which you've talked about with data engine, all the things that we're talking
link |
about with Tesla autopilot, just, uh, transferring that over to the, just yet another robotics problem.
link |
I have to ask, since I care about human robot interaction, so the human side of that,
link |
so you've talked about mostly in the factory. Do you see it, uh, also, do you see part of this
link |
problem that Tesla bot has to solve as interacting with humans and potentially having a place,
link |
like in the home. So interacting, not just not replacing labor, but also like, I don't know,
link |
being a friend or an assistant. Yeah. Yeah. I think the, the possibilities are, you know, endless.
link |
Yeah. I mean, it's, it's, it's obviously like a, it's not quite in Tesla's primary mission
link |
direction of accelerating sustainable energy, but, uh, it is a, an extremely useful thing
link |
that we can do for the world, which is to make a useful humanoid robot. Um, that is capable of
link |
interacting with the world and, um, helping in, in many different ways. Uh, so,
link |
certainly in factories and really just, just, I mean, I think if you say like, uh, extrapolate
link |
to, you know, many years in the future, it's like, I think, uh, work will become optional.
link |
Yeah. So like there's a lot of jobs that if you, if you, if people weren't paid to do it,
link |
they, they wouldn't do it. Like it's not, it's not fun, you know, necessarily. Like
link |
if you're washing dishes all day, it's like, uh, you know, even if you really like washing dishes,
link |
you really want to do it for eight hours a day every day. Probably not. So, um, and then there's
link |
like dangerous work and basically if it's dangerous, boring, uh, it has like potential
link |
for repetitive stress injury, injury, that kind of thing. Um, then that's really where
link |
humanoid robots would add the most value initially. Um, so that's what we're aiming for is, is to, um,
link |
for, for the humanoid robots to do jobs that people don't, don't voluntarily want to do.
link |
Um, and then we'll have to pair that obviously with some kind of universal basic income in the
link |
future. Uh, so I think, um, do you see a world when there's like hundreds of millions of Tesla
link |
bots doing different performing different tasks throughout the world?
link |
Yeah. I haven't really thought about it that far into the future, but I guess that there may be
link |
something like that. Um, so I guess it's a wild question. So the, the number of Tesla cars has
link |
been accelerating. It's been close to 2 million produced. Many of them have autopilot. I think
link |
we're over 2 million now. Yeah. Do you think there will ever be a time when there'll be more Tesla
link |
bots than Tesla cars? Yeah. I, I, I, you know, actually it's funny you ask this question because
link |
normally I do try to think I'm pretty far into the future, but I haven't really thought that far
link |
into the future with the, with the Tesla bot or it's code named Optimus. I call it Optimus subprime
link |
because it's not, it's not like a giant, you know, transformer robot. Um, so, uh,
link |
but it's meant to be a general purpose helpful, helpful bot. Um,
link |
um, and, and basically like the things that we're basically like, like Tesla, I think, um,
link |
is the, has the most advanced real world AI, uh, for interacting with the real world,
link |
which you developed as a function of to, to make self driving work. Um, and so along with custom
link |
hardware and like a lot of, you know, uh, hardcore low level software to have it run efficiently and
link |
be, you know, power efficient because, you know, it's one thing to do neural nets. If you've got a
link |
gigantic solar room with 10,000 computers, but now let's say you just, you have to now distill
link |
that down into one computer that's running at low power in a humanoid robot or a car. Um,
link |
that's actually very difficult and a lot of hardcore software is required for that. Um,
link |
so, so since we're kind of like solving the navigate the real world with neural nets problem
link |
for cars, which are kind of robots with four wheels, then it's like kind of a natural extension
link |
of that is to put it in a robot with arms and legs, uh, and actually, you know, actuators. Um, so,
link |
um, like, like the, the, the two, like hard things are like, you basically need to make the,
link |
have the robot be intelligent enough to interact in a sensible way with the environment. Um,
link |
so you should need real, real world AI and you need to be very good at, um, manufacturing,
link |
which is a very hard problem. Tesla is very good at manufacturing and also has the real world AI.
link |
So making the humanoid robot work is, uh, basically means developing custom, uh, motors and sensors,
link |
uh, that, that are different for what a car would use. Um, but we, we're also, we have a, um,
link |
I think we have the, the, the best expertise in developing advanced electric motors and
link |
power electronics. So it just has to be for a humanoid robot application on a car.
link |
Still, you do talk about love sometimes. So let me ask, this isn't like for like sex robots
link |
or something like that. Love is the answer. Yes. Uh, there is something compelling to us,
link |
not compelling, but we connect with, um, humanoid robots or even legged robots,
link |
like with the dog and shapes of dogs. It just, it seems like, you know, there's a huge amount
link |
of loneliness in this world. All of us seek companionship and with other humans, friendship
link |
and all those kinds of things. We have a lot of here in Austin, a lot of people have dogs.
link |
That's right. Um, there seems to be a huge opportunity to also have robots that decrease
link |
the, uh, the, the amount of loneliness in the world or help us humans connect with each,
link |
with each other. So in the way that dogs can, um, do you think about that?
link |
We'll test about it all. Or is it really focused on the problem of, of performing specific tasks,
link |
not connecting with humans? Um, I mean, to be, to be honest, I have not actually thought about it
link |
from the companionship standpoint, but I think it actually would end up being, it could be actually
link |
a very good companion. Um, and it could, you develop like a personality, uh, over time that is,
link |
that is like unique. Like, uh, you know, it's not like they're just all the robots are the same.
link |
And that personality could evolve to be, you know, uh, match, match the, the, the owner or the,
link |
you know, yes, the owner, uh, well, uh, whatever you want to call it, uh, the other companion,
link |
the other half, right? Uh, in the same way that friends do. See, I think that's a huge opportunity.
link |
I think, yeah, no, that's interesting. Like, um, the, because, you know, like there's, uh,
link |
Japanese phrase, I like the, uh, Wabi Savi, you know, uh, the subtle imperfections
link |
are what makes something special. And the subtle imperfections of the personality of the robot
link |
mapped to the subtle imperfections of the robot's human friend, I don't know,
link |
owner sounds like maybe the wrong word, but, um, could actually make an incredible buddy,
link |
basically. And in that way, the imperfections, like R2D2 or like a C3PO sort of thing, you know.
link |
So from a machine learning perspective, I think the flaws being a feature is really nice.
link |
You could be quite terrible at being a robot for quite a while in the general home environment
link |
or all in the general world. And that's kind of adorable. And that's like, those are your flaws
link |
and you fall in love with those flaws. So it's, it's a, it's very different than autonomous
link |
driving where it's a very high stakes environment. You cannot mess up. And so it's, yeah, it's more
link |
fun to be a robot in the home. Yeah. In fact, if you think of like a C3PO and R2D2, like they
link |
actually had a lot of like flaws and imperfections and silly things and they would argue with each
link |
other. And, um, were they actually good at doing anything? I'm not exactly sure.
link |
I definitely added a lot to the story. Um, but, but, but there's, there's sort of quirky elements
link |
and, you know, that they would like make mistakes and do things. Like it was like, uh, it made them
link |
relatable, I don't know, um, enduring. So, so yeah, I think that that could be something
link |
that probably would happen. Um, but our initial focus is just to make it useful. Uh, so, so,
link |
um, I'm confident we'll get it done. I'm not sure what the exact timeframe is, but uh,
link |
like we'll probably have, I don't know, a decent prototype towards the end of next year or something
link |
like that. And it's cool that it's connected to Tesla, the car. So, so yeah, it's, it's, it's using
link |
a lot of, you know, it would use the autopilot inference computer and, um, a lot of the training
link |
that we've done for the four cars in terms of recognizing real world things could be applied
link |
directly to the, to the robot. Um, so it, but, but there's, there's a lot of custom actuators
link |
and sensors that need to be developed. And an extra module on top of the vector space,
link |
uh, for love. Uh, yeah. That's amazing. Okay. We can add that to the car too.
link |
That's true. Um, that could be useful in all environments. Like you said, a lot of people
link |
argue in the car. So maybe we can help them out. Uh, you're a student of history,
link |
fan of Dan Carlin's hardcore history podcast. Yeah, that's great. Greatest podcast ever.
link |
Yeah. I think it is actually, it almost doesn't really count as a podcast. Yeah. It's more like
link |
a audio book. Yeah. So you were on the podcast with Dan, I just had a chat with him about it.
link |
He said, you guys want military and all that kind of stuff. Uh, yeah, it's literally, uh, it was
link |
basically, um, uh, I think it should be titled engineer wars. Uh, essentially like, like when
link |
there's a rapid change in the rate of technology, then, uh, engineering plays a pivotal role in,
link |
in victory and battle. Um, do you get, how far back in history did you go? Did you go World War
link |
II? Uh, it was mostly, well, it was supposed to be a deep dive on fighters and bomber, uh,
link |
technology in World War II. Um, but that ended up being more wide ranging than that. Um,
link |
because I just went down the, a total rathole of like studying all of the fighters and bombers
link |
of World War II and like the constant rock, paper, scissors game that like, you know, uh,
link |
one country would make this plane, then it'd make a plane to beat that and that's what I'm
link |
trying to make a plane to beat that. And then the, and really what matters is like the pace of
link |
innovation, um, and also access to high quality, uh, fuel and, uh, raw materials. So like Germany
link |
had like some amazing designs, but they couldn't make them, uh, because they couldn't get their
link |
raw materials. Uh, and, uh, they, they had a real problem with the oil and, and, and, uh, fuel
link |
basically the fuel quality was extremely, uh, variable. So the design wasn't the bottleneck
link |
because, uh, yeah, like the US had kick ass fuel, uh, that was like very consistent. Like the
link |
problem is if you make a very high performance aircraft engine, um, in order to make high
link |
performance, you have to, um, the, the, the, the, the fuel, the aviation gas, uh, has to be a consistent
link |
mixture and, uh, uh, it has to have a high octane. Um, like high octane is the most important thing,
link |
but also can't have like impurities and stuff, uh, because you'll, you'll foul up the engine
link |
and, and, and Germany just never had good access to oil. Like they try to get it by invading the
link |
Caucasus, um, but that didn't work too well. Never works well.
link |
That's, that's for you. So there was, Germany was always struggling with, with basically shitty
link |
oil. Um, and then they could not, uh, they, they couldn't count on a, on high quality fuel for
link |
their aircraft. So then they had to have all these additives and stuff. Uh, so, um, uh, whereas the
link |
US had awesome fuel, um, and that provided that to Britain as well. Um, so that allowed the British
link |
and the Americans to design aircraft engines that were, uh, super high performance, better than
link |
anything else in the world. Germany could, could, could design the engines. They just didn't have
link |
the fuel. Uh, and then also the, like I said, the, the, uh, the quality of the aluminum allies that
link |
they were getting was also not that great. And so, you know,
link |
is your, is this like, uh, do you talk about all this with them?
link |
Yeah. Awesome. Broadly looking at history, when you look at Genghis Khan, when you look at Stalin,
link |
Hitler, the darkest moments of human history, uh, what do you take away from those moments?
link |
Does it help you gain insight about human nature, about human behavior today,
link |
whether it's the wars or the individuals or just the behavior of people, any aspects of history?
link |
Yeah, I find history fascinating. Um,
link |
um, there's just a lot of incredible things that have been done, good and bad, um, that they
link |
help, you just help you understand the nature of civilization, um, and individuals and
link |
Does it make you sad that humans do these kinds of things to each other? You look at the 20th
link |
century, World War II, the cruelty, the abuse of power, talk about communism, Marxism and Stalin.
link |
Um, I mean, some of these things do, I mean, if you, like there's a lot of human history,
link |
um, most of it is actually people just getting on with their lives. Uh, you know, and it's not like
link |
human history is just, uh, what nonstop war and disaster is, those are actually just
link |
those are intermittent and rare. And if they weren't, then, you know, humans would soon cease to exist.
link |
Uh, but it's just that wars tend to be written about a lot. And whereas, like, uh,
link |
uh, something being like, well, a normal year where nothing major happened was doesn't get
link |
written about much, but that's, you know, most people just like farming and kind of like living
link |
their life, you know, um, being a villager somewhere. Um, and every now and again, there's a war
link |
and a thing. So, um, and, um, you know, I'd say like that there aren't very many books that I,
link |
where I just had to start reading because it was just too, too dark. But, uh, the book about Stalin,
link |
the court of the red czar, I had to start reading. It was just too, too dark, rough.
link |
Yeah. Um, the thirties, uh, there's a lot, a lot of lessons there to me in particular that it feels
link |
like humans, like all of us have that as the old soldiers in line, um, that the line between good
link |
and evil runs to the heart of every man that all of us are capable of evil. All of us are capable
link |
of good. It's almost like this kind of responsibility that, um, all of us have to, to, to tend towards
link |
the good. And so like to me, looking at history is almost like an example of, look, you have some
link |
charismatic leader that, uh, convinces you of things is too easy based on that story to do evil
link |
onto each other, onto your family and to others. And so it's like our responsibility to do good.
link |
Um, it's not like now is somehow different from history. That can happen again. All of it can
link |
happen again. And yes, most of the time you're right. I mean, the optimistic view here is mostly
link |
people are just living life. And as you've often memed about, uh, the quality of life was way worse
link |
back in the day and keeps improving over time through innovation to technology.
link |
But still it's somehow notable that these blimps of atrocities happen. Sure. Yeah. I mean life was
link |
really tough for most of history. Um, I mean, for most of human history, um, a good year would be
link |
one where not that many people in your village died of the plague, starvation, freezing to death,
link |
or being killed by a neighboring village. It's like, well, it wasn't that bad. You know, it was
link |
only like, you know, we lost 5% this year. That was, uh, it was a good year. You know, that would
link |
be part of the course. Like just, just not starving to death would have been like the primary goal
link |
of most people in through throughout history is making sure we'll have no foods last for the
link |
winter and not get, not freeze or whatever. So, um, now food is, is plentiful. I have an obesity
link |
problem. Um, you know, so. Well, yeah. The lesson there is to be grateful for the way things are
link |
now for, for some of us. We've spoken about this offline. I'd love to get your thought about it here.
link |
If I sat down for a long form in person conversation with the president of Russia,
link |
Vladimir Putin, would you potentially want to call in for a few minutes to join in on a conversation
link |
with a moderated translated by me? Sure. Yeah. Sure. I'd be happy to do that.
link |
You've shown interest in the Russian language. Is this grounded in your interest in history
link |
of linguistics, culture, general curiosity? I think it sounds cool. Sounds cool. Now it looks cool.
link |
Well, it's, it's, you know, it's, it's a, it's, it takes a moment to read Cyrillic.
link |
Once you know what the Cyrillic characters stand for, actually, then reading Russian
link |
becomes a lot easier because there are a lot of words that are actually the same.
link |
Like bank is bank. So find the words that are exactly the same and now you start to understand
link |
Cyrillic. Yeah. If you can, if you can sound it out, it's much, there's at least some commonality
link |
of words. What about the culture? You, you love great engineering, physics. There's a tradition
link |
of the sciences there. Sure. You look at the 20th century from rocketry. So, you know,
link |
some of the greatest rockets of the space exploration has been done in the Soviet and the
link |
former Soviet Union. Yeah. So do you draw inspiration from that history? Just how this
link |
culture that in many ways, I mean, one of the sad things is because of the language,
link |
which a lot of it is lost to history because it's not translated to all those kinds of,
link |
because it, it is in some ways an isolated culture. It flourishes within its, within its borders.
link |
Yeah. So do you draw inspiration from those folks from, from the history of science engineering
link |
there? I mean, the Soviet Union, Russia, and Ukraine as well, and have a really strong history
link |
in spaceflight. Like some of the most advanced and impressive things in history were done,
link |
you know, by the Soviet Union. So one can, cannot help but admire the
link |
impressive rocket technology that was developed. You know, after the sort of full Soviet Union,
link |
the, there's, there's much less that, that, than happened. But
link |
still things are happening, but it's not, not quite at the
link |
frenetic pace that was happening before the Soviet Union kind of
link |
dissolved into separate republics. Yeah. I mean, I, I, you know, there's Roscosmos,
link |
the Russian agency. I look forward to a time when those countries with China are working together,
link |
the United States are all working together. Maybe a little bit of friendly competition, but
link |
like friendly competition is good. You know, government's so slow, and the only thing slower
link |
than one government is a collection of governments. So the Olympics would be boring if everyone just
link |
crossed the finishing line at the same time. Yeah. Nobody would watch. Yeah. And, and people wouldn't
link |
try hard to run fast and stuff. So I think friendly competition is a good thing.
link |
This is also a good place to give a shout out to a video titled the entire Soviet rocket engine
link |
family tree by Tim Dodd, aka everyday astronaut. It's like an hour and a half. It gives a full
link |
history of Soviet rockets. And people should definitely go check on support Tim in general.
link |
That guy's super excited about the future, super excited about a spaceflight. Every time I see
link |
anything by him, I just have a stupid smile on my face because he's so excited about stuff.
link |
Yeah. Tim Dodd is really, really great. If you're interested in anything to do with space,
link |
he's, in terms of explaining rocket technology to your average person, he's awesome. The best,
link |
I'd say. And I should say like the part of the reason like I switched us from, like Raptor at
link |
one point was going to be a hydrogen engine. But hydrogen has a lot of challenges. It's very low
link |
density. It's a deep cryogen. So it's only liquid at a very, you know, very close to absolute zero
link |
requires a lot of insulation. It's, so it is a lot of challenges there.
link |
And I was actually reading a bit about Russian rocket engine developments. And
link |
at least the impression I had was that Soviet Union Russia and Ukraine primarily were
link |
actually in the process of switching to methalox. And there was some interesting
link |
test and data for ISP, like they were able to get like up to like a 382nd ISP with the
link |
methalox engine. And I was like, well, okay, that's, that's actually really impressive. So
link |
so I think we could, you could actually get a much lower cost, like in optimizing cost per
link |
time to orbit cost per time to Mars. It's, I think methane oxygen is the way to go.
link |
And I was partly inspired by the Russian work on the test stands with methalox engines.
link |
And now for something completely different. Do you mind doing a bit of a meme review in the
link |
spirit of the great, the powerful beauty pie, let's say one to 11, just go over a few documents
link |
printed out. We can try. Let's try this. I present to you document number Uno.
link |
I don't know. Okay. Vlad the Impaler discovers marshmallows.
link |
Yeah, that's not bad.
link |
So you get it because he's failing things. I don't know, three, whatever.
link |
That's not very good.
link |
This is grounded in some engineering, some history.
link |
Yeah, give us an eight out of 10.
link |
What do you think about nuclear power?
link |
I'm in favor of nuclear power. I think it's in a place that is not subject to extreme natural
link |
disasters. I think it's a nuclear power is a great way to generate electricity.
link |
I don't think we should be shutting down nuclear power stations.
link |
Yeah, but what about Chernobyl?
link |
Exactly. So I think people, there's like a lot of fear of radiation and stuff.
link |
And it's, I guess, probably like a lot of people just don't
link |
even study engineering or physics. It's just the word radiation just sounds scary.
link |
You know, so they don't, they can't calibrate what radiation means. But radiation is much
link |
less dangerous than you think. So like, for example, Fukushima, when the Fukushima problem
link |
happened due to the tsunami, I got people in California asking me if they should worry about
link |
radiation from Fukushima. And I'm like, definitely not, not even slightly, not at all. That is crazy.
link |
And just to show, like, look, this is how, like, the danger is so much overplayed compared to what
link |
it really is that I actually flew to Fukushima and I donated a solar power system for what
link |
a treatment plant. And, and I made a point of eating locally grown vegetables on TV in Fukushima.
link |
Like, I'm still alive. Okay.
link |
So it's not even at the risk of these events as low, but the impact of them is.
link |
Impact is greatly exaggerated. It's just human nature.
link |
It's people who don't know what radiation is, like I've had people ask me, like, what about
link |
radiation from cell phones, quoting, causing brain cancer? I'm like, when you say radiation,
link |
do you mean photons or particles than like that? I don't know what, what do you mean
link |
photons particles? So do you mean, let's say photons? What, what, what frequency,
link |
wavelength? And they're like, no idea. Like, do you know that everything's radiating all the time?
link |
Like, what do you mean? Like, yeah, everything's radiating all the time.
link |
Photons are being emitted by, by all objects all the time, basically. So,
link |
and if you want to know what it's, it's what, what it means to stand in front of nuclear fire,
link |
go outside. The sun is a gigantic, you know, thermonuclear reactor that you're staring
link |
right at it. Are you still alive? Yes. Okay. Amazing.
link |
Yeah. I guess radiation is one of the words that could be used as a tool to, to, to fear
link |
monger by certain people. That's it. And I think people just don't, don't understand. So, I mean,
link |
that's the way to fight that, that fear, I suppose, is to understand, is to learn.
link |
Yeah. Just say like, okay, how many people have actually died from nuclear accidents? It's like
link |
practically nothing. And say how many people have, have died from, you know, coal plants? And
link |
it's a very big number. So like, obviously, we should not be starting up coal plants and shutting
link |
down nuclear plants. It just doesn't make any sense at all. Coal plants, like, I don't know,
link |
a hundred to a thousand times worse for, for health than nuclear power plants.
link |
You want to go to the next one? This is really bad.
link |
So that 90, 180 and 360 degrees, everybody loves the math, nobody gives a shit about 270.
link |
It's not super funny. I don't like 203. Yeah. This is not, you know, LOL situation.
link |
Yeah. That's pretty good.
link |
The United States oscillating between establishing and destroying dictatorships.
link |
It's like a metro. Is that a metro? Yeah. What does that mean? Yeah.
link |
Yeah. It's a 7 out of 10. It's kind of true. Oh, yeah. This is,
link |
this is kind of personal for me. Next one. Oh, man. This is Leica?
link |
Yeah. Well, no, this is... Or it's like referring to Leica or something?
link |
As Leica is like a husband. Hello. Yes. This is dog. Your wife was launched to space.
link |
And then the last one is him with his eyes closed and a bottle of vodka.
link |
Yeah. Leica didn't come back. No. They don't tell you the full story of, you know,
link |
what the love, the impact they had on the loved ones.
link |
True. That one gets an 11 for me. Sure.
link |
The Soviet set up. Yeah. This keeps going on the Russian theme. First man in space,
link |
nobody cares. First man on the moon. Well, I think people do care.
link |
No, I know. But...
link |
Yuri Gagarin's names will be forever in history, I think.
link |
There is something special about placing like stepping foot onto another totally foreign land.
link |
It's not the journey like people that explore the oceans. It's not as important to explore the
link |
oceans as to land on a whole new continent. Yeah.
link |
Well, this is about you. Oh, yeah. I'd love to get your comment on this.
link |
Elon Musk, after sending $6.6 billion to the UN to end world hunger, you have three hours.
link |
Yeah. Well, I mean, obviously $6 billion is not going to end world hunger. So
link |
so I mean, the reality is at this point, the world is producing
link |
far more food than it can really consume it. Like we don't have a caloric
link |
constraint to this point. So where there is hunger, it is almost always due to
link |
like civil war, strife or some like... It's not a thing that is
link |
extremely rare for it to be just a matter of lack of money. It's like,
link |
you know, it's like some, the civil war in some country and like one part of the country is
link |
literally trying to starve the other part of the country. So it's much more complex than
link |
something that money could solve. It's geopolitics. It's a lot of things. It's human nature.
link |
It's governments. It's money, monetary systems, all that kind of stuff.
link |
Yeah. Food is extremely cheap these days. It's like,
link |
I mean, the US at this point, you know, among low income families, obesity is actually another
link |
problem. It's not, like obviously it's not hunger. It's like too many calories. So
link |
it's not that nobody's hungry anywhere. It's just, this is not a simple matter of adding money and
link |
solving it. What do you think that one gets? It's getting... Two.
link |
Just going after Empire's world. Where did you get those artifacts? The British Museum
link |
has a shout out to Monty Python. We found them.
link |
Yeah. The British Museum is pretty great. I mean,
link |
immediately Britain did take these historical artifacts from all around the world and put
link |
them in London. But, you know, it's not like people can't go see them. So it is a convenient
link |
place to see these ancient artifacts is London for, you know, for a large segment of the world.
link |
So I think, you know, on balance, the British Museum is a net good. Although I'm sure
link |
that a lot of countries argue about that. Yeah. It's like you want to make these historical
link |
artifacts accessible to as many people as possible. And the British Museum, I think,
link |
does a good job of that. Even if there's a darker aspect to, like, the history of Empire in general,
link |
whatever the Empire is, however things were done, it is the history that happened. You
link |
can't sort of erase that history, unfortunately. You could just become better in the future.
link |
That's the point. Yeah. I mean, it's like, well, how are we going to pass from all judgment on
link |
these things? Like, it's like, you know, if one is going to judge, say, the British Empire,
link |
you got to judge, you know, what everyone was doing at the time, and how were the British
link |
relative to everyone? And I think they were first would actually get like a relatively good grade,
link |
relatively good grade, not an absolute terms, but compared to what everyone else was doing.
link |
They were not the worst. Like I said, you got to look at these things in the context of the
link |
history at the time, and say, what were the alternatives? And what are you comparing it
link |
against? Yes. And I do not think it would be the case that Britain would get a bad grade
link |
in when looking at history at the time. You know, if you judge history from, you know,
link |
from what is morally acceptable today, you basically are going to give everyone a failing
link |
grade. I'm not clear. I don't think anyone would get a passing grade in their morality of,
link |
like you go back 300 years ago, like, who's getting a passing grade? Basically, no one.
link |
And we might not get a passing grade from generations that come after us. What does that one get?
link |
Sure. Six, seven, seven.
link |
For the Monty Python, maybe. I always love Monty Python. They're great.
link |
I like for Brian and the Quist of the Holy Grail are incredible.
link |
Yeah, yeah. Yeah, those serious eyebrows. Brezhnev. How important do you think is facial hair to
link |
great leadership? Well, you got a new haircut. How does that affect your leadership?
link |
I don't know. Hopefully not. It doesn't.
link |
Yeah, the second is no one.
link |
There is no one competing with Brezhnev. No one, too.
link |
Those are like epic eyebrows. So, sure. That's ridiculous.
link |
Give it a six or seven. I don't know. I like this, like, Shakespeare analysis of memes.
link |
Brezhnev, he had a flair for drama as well. Like, you know, showmanship.
link |
Yeah, yeah. It must come from the eyebrows. All right. Invention, great engineering.
link |
Look what I invented. That's the best thing since ripped up bread.
link |
Yeah. Because they invented sliced bread. Am I just explaining memes at this point?
link |
This is what my life has become.
link |
You're going to be more of a meme explainer.
link |
Yeah, I'm a meme. Like a scribe that like runs around with the kings and just like writes down
link |
memes. I mean, when was the cheeseburger invented? That's like an epic invention.
link |
You know, versus just like a burger?
link |
Or burger. I guess a burger in general. It's like, you know.
link |
Then there's like, what is a burger? What's the sandwich? And then you start getting
link |
the pizza sandwich and what is the original? It gets into an ontology argument.
link |
Yeah. But everybody knows like, if you order like a burger or cheeseburger or whatever and you
link |
like, you got like, you know, tomato and some lettuce and onions and whatever and, you know,
link |
you know, mayo and ketchup and mustard. It's like epic.
link |
Yeah. But I'm sure they've had bread and meat separately for a long time and it was kind of
link |
a burger on the same plate, but somebody who actually combined them into the same thing
link |
and you buy it and hold it, makes it convenient. It's a materials problem. Like your hands don't
link |
get dirty and whatever. Yeah, it's brilliant.
link |
Well, that is not what I would have guessed.
link |
But everybody knows like, if you order a cheeseburger, you know what you're getting,
link |
you know, it's not like some obtuse. Like, I wonder what I'll get, you know.
link |
You know, fries are, I mean, great. I mean, they're the devil, but fries are awesome. And
link |
yeah, pizza is incredible. Food innovation doesn't get enough love.
link |
Yeah. I guess is what we're getting at.
link |
It's great. What about the Matthew McGonaghey Austinite here?
link |
President Kennedy, do you know how to put men on the moon yet?
link |
NASA know. President Kennedy, be a lot cooler if you did.
link |
Pretty much. Sure. Six, six or seven.
link |
And this is the last one. That's funny.
link |
Someone drew a bunch of dicks all over the walls of Sistine Chapel Boys Bathroom.
link |
Sure. I'll give it nine. It's really true.
link |
This is our highest ranking meme for today.
link |
I mean, it's true. Like, how do they get away with that?
link |
Lots of nakedness. I mean, dick pics are, I mean, just something throughout history.
link |
As long as people can draw things, there's been a dick pic.
link |
It's a staple of human history.
link |
It's a staple. Consistence throughout human history.
link |
You tweeted that you aspire to comedy. Your friends with Joe Rogan
link |
might you do a short stand up comedy set at some point in the future?
link |
Maybe open for Joe, something like that. Is that really stand up?
link |
Actual just full on stand up? Full on stand up. Is that in there or is that?
link |
I've never thought about that.
link |
It's extremely difficult if at least that's what Joe says in the comedians say.
link |
Huh. I wonder if I could. The only one way to find out.
link |
You know, I have done stand up for friends just impromptu.
link |
You know, I'll get on like a roof and they do laugh, but they're our friends too.
link |
So I don't know if you've got to call, you know, like a room of strangers.
link |
Are they going to actually also find it funny? But I could try. See what happens.
link |
I think you'd learn something either way. Yeah, I kind of love
link |
both the, when you bomb and when, when you do great just watching people,
link |
how they deal with it. It's so difficult. It's so, you're so fragile
link |
up there. It's just you and you think you're going to be funny.
link |
And when it completely falls flat, it's just, it's beautiful to see people
link |
deal with like that. I might have enough material to do stand up.
link |
I've never thought about it, but I might have enough material.
link |
I don't know, like 15 minutes or something. Oh yeah. Yeah. Do a Netflix special.
link |
Netflix special. Sure. What's your favorite Rick and Morty concept?
link |
Just to spring that on you. Is there, there's a lot of sort of scientific engineering ideas
link |
explored there. There's the favorite. There's the butter robot. It's great.
link |
Yeah. It's a great show. You like it? Yeah. Rick and Morty is awesome.
link |
Somebody that's exactly like you from an alternate dimension showed up there, Elon Tusk.
link |
Yeah. That's right. That you voiced.
link |
Yeah. Rick and Morty certainly explores a lot of interesting concepts.
link |
Like what's the favorite one? I don't know. The butter robot certainly is,
link |
you know, it's like, it's certainly possible to have too much sentence in a device.
link |
Like you don't want to have your toast to be like a super genius toaster.
link |
It's going to hate life because all it could just make is toast. But if, you know, it's like,
link |
you don't want to have like super intelligence stuck in a very limited device.
link |
Do you think it's too easy from a, if we're talking about from the engineering perspective
link |
of super intelligence, like with Marvin, the robot, like is it just, it seems like it might
link |
be very easy to engineer just a depressed robot. Like it, it's not obvious to engineer a robot
link |
that's going to find a fulfilling existence. Same as humans, I suppose. But I wonder if that's like
link |
the default. If you don't do a good job on building a robot, it's going to be sad a lot.
link |
Well, we can reprogram robots easier than we can reprogram humans. So I guess if you let it evolve
link |
without tinkering, then it might get sad. But you can change the optimization function and
link |
have it be a cheery robot.
link |
You, like I mentioned with, with SpaceX, you give a lot of people hope. And a lot of people look
link |
up to you, millions of people look up to you. If we think about young people in high school,
link |
maybe in college, what advice would you give to them about if they want to try to do something
link |
big in this world, they want to really have a big positive impact, what advice would you give them
link |
about their career, maybe about life in general?
link |
Try to be useful. You do things that are useful to your fellow human beings to the world. It's
link |
very hard to be useful. Very hard. You know, are you contributing more than you consume?
link |
You know, like, like, can you try to have a positive net contribution to society?
link |
I think that's the thing to aim for, you know, not to try to be sort of a leader for,
link |
for the sake of being a leader or whatever. A lot of time people who, a lot of time, the people
link |
you want as leaders are the people who don't want to be leaders. So if you can live a useful life,
link |
that is a good life, a life worth having lived. You know, and like I said, I would encourage people
link |
to use the mental tools of physics and apply them broadly in life. They are the best tools.
link |
When you think about education and self education, what do you recommend? So there's the university,
link |
there's self study, there is a hands on sort of finding a company or a place or a set of people
link |
that do the thing you're passionate about and joining them as early as possible.
link |
There's taking a road trip across Europe for a few years and writing some poetry,
link |
which trajectory do you suggest in terms of learning about how you can become useful,
link |
as you mentioned, how you can have the most positive impact?
link |
Well, I encourage people to read a lot of books. Basically, try to ingest as much information as
link |
you can and try to also just develop a good general knowledge. So you at least have a rough
link |
lay of the land of the knowledge landscape. Try to learn a little bit about a lot of things.
link |
Because you might not know what you're really interested in. How would you know what you're
link |
really interested in if you at least aren't doing it peripheral exploration or broadly of
link |
the knowledge landscape? And you talk to people from different walks of life and different
link |
industries and professions and skills and occupations. Just try to learn as much as possible.
link |
Man's search for meaning.
link |
Isn't the whole thing a search for meaning?
link |
Yeah, what's the meaning of life and all? But just generally, like I said, I would encourage
link |
people to read broadly in many different subject areas and then try to find something where there's
link |
an overlap of your talents and what you're interested in. So people may be good at something,
link |
but they may have skill at a particular thing, but they don't like doing it.
link |
So you want to try to find a thing where that's a good combination of the things that you're
link |
inherently good at, but you also like doing. And reading is a super fast shortcut to
link |
figure out where are you. You're both good at it. You like doing it, and it will actually have positive
link |
impact. Well, you got to learn about things somehow. So reading a broad range, it's just
link |
really read it. One important one is that kid I read through the encyclopedia. So that's pretty
link |
helpful. And those are things I didn't even know existed. Well, lots, obviously. It's like as
link |
broad as it gets. Encyclophilias were digestible, I think, whatever, 40 years ago. So maybe read
link |
through the condensed version of the encyclopedia Britannica. I'd recommend that. You can always
link |
like skip subjects or you read a few paragraphs, and no, you're not interested. Just jump to the
link |
next one. So read the encyclopedia or scan through it. And you know, put a lot of stock and
link |
certainly have a lot of respect for someone who puts in an honest day's work to do useful things.
link |
And just generally to have like not a zero sum mindset, or like have more of a
link |
like, grow the pie mindset, like the, if you sort of say like, when we see people like,
link |
perhaps, including some very smart people, kind of taking an attitude of like, like,
link |
like doing things that seem like morally questionable. It's often because they have
link |
at a base sort of axiomatic level, a zero sum mindset. And, and they without realizing it,
link |
they don't realize they have a zero sum mindset, or at least they don't realize it consciously.
link |
And so if you have a zero sum mindset, then the only way to get ahead is by taking things from
link |
others. If it's like, if the, if the, if the pie is fixed, then the only way to have more pie is to
link |
take someone else's pie. But, but this is false, like obviously the pie has grown dramatically
link |
over time, the economic pie. So the reality, in reality, you can have, so over useless analogy,
link |
if you have a lot of, there's a lot of pie. Pie pie is not fixed. So you really want to make sure
link |
you don't, you're not operating without realizing it from a zero sum mindset, where, where the only
link |
way to get ahead is to take things from others, then that's going to result in you trying to
link |
take things from others, which is not, not good. It's much better to work on adding to the economic
link |
pie, maybe, you know, so creating, like I said, creating more than you consume, doing more than
link |
you. Yeah. So that's a big deal. I think there's like, you know, a fair number of people in,
link |
in finance that do have a bit of a zero sum mindset.
link |
I mean, it's all walks of life. I've seen that. And one of the, one of the reasons
link |
Rogan inspires me is he celebrates others a lot. This is not, not creating a constant competition.
link |
Like there's a scarcity of resources. What happens when you celebrate others and you promote others,
link |
the ideas of others, it, it, it actually grows that pie. I mean, it, every, like the,
link |
the resource, the resources become less scarce. And that, that applies in a lot of kinds of
link |
domains. It applies in academia where a lot of people are very, see some funding for academic
link |
research is zero sum. And it is not, if you celebrate each other, if you make, if you get
link |
everybody to be excited about AI, about physics, about mathematics, I think it, there'll be more,
link |
more funding. And I think everybody wins. Yeah. That applies, I think broadly.
link |
Yeah. Yeah. Exactly. So last, last, last question about love and meaning.
link |
What is the role of love in the human condition broadly and more specific to you? How has love,
link |
romantic love, or otherwise made you a better person, a better human being?
link |
Now you're asking really perplexing questions. It's hard to give up. I mean, there are many
link |
books, poems, and songs written about what is love and what is, what exactly, you know,
link |
you know, what is love? Maybe you don't hurt me.
link |
That's one of the great ones. Yes. Yeah. You've, you've earlier quoted Shakespeare,
link |
but that, that's really up there. Yeah. Love is a many splendid thing.
link |
I mean, there's, because we've talked about so many inspiring things like be useful in the world,
link |
sort of like solve problems, alleviate suffering, but it seems like connection between humans is a
link |
source, you know, it's a source of joy is a source of meaning. And that, that's what love is, friendship,
link |
love. I just wonder if you think about that kind of thing when you talk about preserving the light
link |
of human consciousness or it's becoming a multi planetary multi planetary species.
link |
I mean, to me, at least that, that means like, if we're just alone and conscious and intelligent,
link |
it doesn't mean nearly as much as if we're with others, right? And there's some magic created
link |
when we're together. The, the French of it, and I think the highest form of it is love,
link |
which I think broadly is, is much bigger than just sort of romantic, but also yes,
link |
romantic love and family and those kinds of things.
link |
Well, I mean, the reason I guess I care about us becoming multi planetary species in a space
link |
frank civilization is foundationally, I love humanity. And, and so I wish to see it prosper and
link |
do great things and be happy. And if I did not love humanity, I would not care about these things.
link |
So when you look at the whole of it, the human history, all the people who's ever lived, all
link |
the people alive now, it's pretty, we're okay. On the whole, we're pretty interesting bunch.
link |
Yeah. All things considered. And I've read a lot of history, including the darkest,
link |
worst parts of it. And despite all that, I think on balance, I still love humanity.
link |
You joked about it with the 42. What do you think is the meaning of this whole thing?
link |
Is it like, is there a non numerical representation?
link |
Yeah. Well, really, I think what Dr. Sattons was saying in Hitchhiker's Guide to the Galaxy is that
link |
the universe is the answer. And what we really need to figure out are what questions to ask
link |
about the answer that is the universe. And that the question is the really the hard part. And
link |
if you can properly frame the question, then the answer relatively speaking is easy.
link |
So therefore, if you want to understand what questions to ask about the universe,
link |
you want to understand the meaning of life. We need to expand the scope and scale of consciousness
link |
so that we're better able to understand the nature of the universe and understand the
link |
meaning of life. And ultimately, the most important part would be to ask the right question.
link |
Yes. Thereby elevating the role of the interviewer.
link |
Yes, exactly. As the most important human in the room.
link |
Good questions are, it's hard to come up with good questions. Absolutely.
link |
But yeah, it's like that is the foundation of my philosophy is that I am curious about the
link |
nature of the universe. And obviously, I will die. I don't know when I'll die, but I won't live
link |
forever. But I would like to know that we are on a path to understanding the nature of the universe
link |
and the meaning of life and what questions to ask about the answer that is the universe.
link |
And so if we expand the scope and scale of humanity and consciousness in general,
link |
which includes Silicon Consciousness, then that seems like a fundamentally good thing.
link |
Elon, like I said, I'm deeply grateful that you have spent your extremely valuable time with me
link |
today and also that you have given millions of people hope in this difficult time, this divisive
link |
time in this cynical time. So I hope you do continue doing what you're doing. Thank you
link |
so much for talking today. You're welcome. Thanks for excellent questions.
link |
Thanks for listening to this conversation with Elon Musk. To support this podcast,
link |
please check out our sponsors in the description. And now let me leave you with some words from
link |
Elon Musk himself. When something is important enough, you do it, even if the odds are not
link |
in your favor. Thank you for listening and hope to see you next time.